Vehicle edge calculation predictive task unloading method and system and storage medium

Abstract

The invention belongs to the technical field of Internet of Vehicles, and particularly relates to a vehicle edge calculation predictive task unloading method and system and a storage medium. The invention discloses a digital twin auxiliary vehicle edge calculation predictive task unloading method. The method comprises the following steps: sensing all task unloading requirements in a domain; receiving the available computing resource state from the edge vehicle and forwarding the available computing resource state to the digital twin module; performing parameter simulation through a digital twin module, and judging whether the fog node accepts the task unloading demand or not according to a parameter simulation result and an actual resource stock to obtain a task unloading decision; and for the accepted task unloading decision, the edge vehicle unloads the task to the fog node for task execution, and then an output result is sent back to the edge vehicle. The invention also provides a system for realizing the method. According to the method, task completion delay can be reduced, fog node resource allocation balance is improved, virtual and real errors are adapted, the calculation rate of a VEC system is improved, and the method has a good application prospect.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle networking, and specifically relates to a vehicle edge computing predictive task offloading method, system and storage medium. Background Art

[0002] The advancement of IoT and wireless communication technologies has driven the explosive growth of smart terminal devices and data traffic. It is expected that in the 6G era, millions of terminal devices will be connected to wireless networks, and mobile applications such as online games, virtual / augmented reality, and autonomous driving will generate a large amount of data at the edge of the wireless network. This data is transmitted to the core network through technologies such as wireless communication, GPS, and sensors, bringing unprecedented challenges to the existing wireless network architecture.

[0003] Edge computing effectively improves the performance of services and applications by migrating computing and caching functions to the edge of the network. However, due to the limited resources of a single edge cloud or terminal device, multiple edge nodes or core clouds are required to collaborate to complete tasks. The vehicle edge computing (VEC) network is a concept that combines mobile edge computing (MEC) and Internet of Vehicles (IoV) technologies. It solves the problem of limited computing power of vehicle terminals by sinking on-board tasks to the edge of the network. The VEC network aims to reduce the latency and energy consumption of executing terminal tasks and improve service quality and user experience quality by allocating tasks between vehicles and edge servers.

[0004] In recent years, as an important means of integrating physics and information, digital twins have attracted widespread attention from various industries for their characteristics of synchronous mapping of virtual and real, symbiosis and integration, and have provided a feasible solution to solve the dynamic complexity of VEC networks. Digital twins establish precise mapping with VEC networks, monitor their dynamic changes in real time, capture the characteristics of service demand and resource utilization, and realize the function of efficient intelligent decision-making assistance. At present, some studies have attempted this, such as the literature "IEEE Transactions on Intelligent Vehicles, 2022, 2(3):212-218.", "IEEE Internet of Things Journal, 2022, 6(04): 125-133.", "IEEE Transactions on Industrial Informatics, 2021, 18(2):1405-1413.", "IEEE Communications Magazine, 2019, 57(5):64-69.", etc.

[0005] Although simulation applications based on digital twins have achieved certain research results, there is still room for improvement in the dynamic mapping of simulation models. In the actual environment of vehicle edge computing, digital twin technology provides the simulation model with the ability to dynamically update, and can cope with disturbances during operation, thereby maintaining virtual-real consistency. However, in actual engineering applications, in order to meet the algorithm's high requirements for the diversity of disturbance types and recognition accuracy, complex algorithms and a large amount of operating data are usually required. Digital twin systems need to solve the above problems to ensure the accuracy and timeliness of simulation results. Therefore, this field urgently needs to develop new task offloading strategies to more reasonably select tasks to be performed by terminal devices, edge computing servers or cloud computing servers in real time, minimize system latency and resource consumption, and achieve wireless network resource consumption and node resource allocation rationalization. Summary of the invention

[0006] In response to the problems of the prior art, the present invention provides a digital twin-assisted vehicle edge computing predictive task offloading method and system.

[0007] A digital twin-assisted vehicle edge computing predictive task offloading method performs the following steps on the fog node: Step 1: Sense all task offloading requirements in the domain; Step 2: Receive the available computing resource status from the edge vehicle and forward it to the digital twin module; Step 3: Perform parameter simulation through the digital twin module, and judge whether the fog node accepts the task offloading demand based on the parameter simulation results and the actual resource inventory, and obtain the task offloading decision; Step 4: For the accepted task offloading decision, the edge vehicle offloads the task to the fog node for task execution, and then sends the output result back to the edge vehicle.

[0008] Preferably, in step 1, the fog node is responsible for maintaining a list of edge vehicles that access and leave the fog node's communicable network; when an edge vehicle accesses, it is initialized and sends its task offloading requirements to the fog node.

[0009] Preferably, in step 3, the digital twin module is configured to: create a network topology of all connected nodes and task interaction relationships, use a digital twin model to perform parameter simulation on the task offloading decisions and data transmission methods of actual fog nodes or edge nodes, and realize the prediction of task computing capabilities of fog nodes.

[0010] Preferably, step 3 specifically includes the following steps: Step 3.1: Input the historical task offloading data generated in the domain into the AI ​​model integrated in the digital twin model; Step 3.2: Calculate the computing time required to execute the vehicle task at the next moment based on the vehicle movement pattern, available resource status and the amount of historical task offloading data; Step 3.3: Compare the calculation time of the vehicle task at the next moment and the task waiting delay contained in the task offloading requirement. If the calculation time of the vehicle task at the next moment is lower than the task waiting delay, the fog node accepts the task offloading requirement, otherwise it rejects it.

[0011] Preferably, in step 3.2, the artificial intelligence model algorithm is selected from LSTM, RNN, CNN, Bayesian method or ARIMA time series model.

[0012] Preferably, in step 3.3, the criteria for determining whether the fog node accepts the task offloading decision also include: when the edge vehicle is directly connected to the fog node and the fog node is executing the task, the task offloading decision can be accepted, otherwise it is rejected.

[0013] Preferably, step 4 specifically includes the following steps: Step 4.1: When the task offloading decision is accepted, the edge vehicle offloads the task to the fog node, and all accepted tasks are saved in the local queue of the fog node until the resources meet the needs of stable task execution; Step 4.2: After the task is executed, all output results are stored in the result queue of the fog node; Step 4.3: The fog node’s scheduler sends the output results back to the edge vehicle.

[0014] Preferably, the following steps are performed locally on the edge vehicle: Step a: The edge vehicle generates a task and stores it in a local queue; Step b: The resource manager of the edge vehicle perceives the decision basis of task offloading in real time, calculates the task requirements and the current workload of the device, and decides whether the task is processed locally or by the fog node. For the task processed by the fog node, the task offloading requirement is obtained; the decision basis of task offloading includes vehicle metadata, task data, fog resources and task deadline information; Step c: temporarily storing the task offloading demand in the offloading queue; Step d: When the edge vehicle is connected to the fog node, it sends the task offloading requirement.

[0015] The present invention also provides a system for implementing the above-mentioned digital twin assisted vehicle edge computing predictive task offloading method, comprising: a fog node and at least one edge vehicle; the fog node is connected to a digital twin module.

[0016] The present invention also provides a computer-readable storage medium, on which is stored: a computer program for implementing the above-mentioned digital twin-assisted vehicle edge computing predictive task offloading method, or a computer program for implementing the above-mentioned system.

[0017] In the present invention, the "task" is the data that needs to be calculated; the "task offloading" places the data that needs to be calculated in the fog node or locally for processing; the "task offloading decision" is the decision result of placing the data that needs to be calculated in the fog or locally for processing.

[0018] The present invention proposes a digital twin-assisted vehicle edge computing predictive task offloading framework, which realizes the digital twin mapping between edge vehicles and fog nodes, and estimates, predicts and evaluates the real-time status of the VEC network by creating virtual copies. A method for node resource allocation based on the prediction results of digital twin parameters is provided, which can optimize the utilization of edge devices, minimize task completion delays and achieve balanced task offloading between fog nodes. Through simulation verification, it is found that the scheme proposed in this article is superior to other existing frameworks in reducing task completion delays, improving the balance of fog node resource allocation, adapting to virtual-real errors and improving the calculation rate of the VEC system. Therefore, the present invention has a good application prospect.

[0019] Obviously, according to the above contents of the present invention, in accordance with common technical knowledge and customary means in the art, without departing from the above basic technical ideas of the present invention, other various forms of modification, replacement or change may be made.

[0020] The above contents of the present invention are further described in detail below through specific implementation methods in the form of embodiments. However, this should not be understood as the scope of the above subject matter of the present invention being limited to the following examples. All technologies realized based on the above contents of the present invention belong to the scope of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a schematic diagram of the system composition of the present invention. DETAILED DESCRIPTION

[0022] It should be noted that the algorithms of data collection, transmission, storage and processing steps not specifically described in the embodiments, as well as the hardware structure, circuit connection, etc. not specifically described can all be implemented through the contents disclosed in the prior art.

[0023] Embodiment 1 Digital twin assisted vehicle edge computing predictive task offloading method and system This embodiment provides a vehicle edge computing (VEC) network capable of executing a digital twin-assisted vehicle edge computing predictive task offloading method.

[0024] Specifically, the system Figure 1As shown, it includes: interconnected fog nodes and edge vehicles (autonomous driving vehicles). Among them, the edge vehicle collects data through sensors, uploads it to the fog node, and uses the data to create a virtual copy to estimate, predict and evaluate the real-time status of VEC. A digital twin module is integrated in the fog node. The digital twin module is configured to: create a network topology of all connected nodes (including node speed and direction, road topology) and task interaction relationships (including task timeliness information, data scale, and computing time), and use the digital twin model to perform parameter simulation (including resources, task deadlines, vehicle mobility, network topology and other parameters) on the task offloading decision and data transmission method of the actual fog node or edge node to achieve prediction of the task computing capacity of the fog node.

[0025] The method for offloading predictive tasks by edge computing of digital twin-assisted vehicles is carried out through the above system. The specific steps are as follows: The following steps are performed locally on the edge vehicle: Step a: The edge vehicle generates a task and stores it in a local queue; Step b: The resource manager of the edge vehicle perceives the decision basis of task offloading in real time, calculates the task requirements and the current workload of the device, and decides whether the task is processed locally or by the fog node. For the task processed by the fog node, the task offloading requirement is obtained; the decision basis of task offloading includes vehicle metadata, task data, fog resources and task deadline information; Step c: temporarily storing the task offloading demand in the offloading queue; Step d: When the edge vehicle is connected to the fog node, it sends the task offloading requirement.

[0026] Perform the following steps on the fog node: Step 1: Sense all task offloading requirements in the domain. In this step, the fog node is responsible for maintaining a list of edge vehicles that access and leave the fog node’s communication network. When an edge vehicle accesses, it is initialized and sends its task offloading requirements to the fog node. Step 2: Receive the available computing resource status from the edge vehicle and forward it to the digital twin module; Step 3: Perform parameter simulation through the digital twin module, and judge whether the fog node accepts the task offloading demand based on the parameter simulation results and the actual resource inventory, and obtain the task offloading decision; Step 4: For the accepted task offloading decision, the edge vehicle offloads the task to the fog node for task execution, and then sends the output result back to the edge vehicle.

[0027] Step 3 specifically includes the following steps: Step 3.1: Input the historical task offloading data generated in the domain into the AI ​​model integrated in the digital twin model; Step 3.2: Calculate the computing time required to execute the vehicle task at the next moment based on the vehicle movement pattern, available resource status and the amount of historical task offloading data; Step 3.3: Determine whether the fog node accepts the task offloading decision. The criteria include: 1) Compare the calculation time of the vehicle task at the next moment with the task waiting delay contained in the task offloading requirement. If the calculation time of the vehicle task at the next moment is lower than the task waiting delay, the fog node accepts the task offloading requirement, otherwise it rejects it. 2) When the edge vehicle is directly connected to the fog node and the fog node is executing a task, it can accept the task offloading decision, otherwise it will reject it.

[0028] The following is the basic definition of the vehicle task offloading model, which is a modeling representation of step 3.3: The vehicle's fog / edge collaborative computing mode is based on the system's calculation time for the vehicle to unload the task at the next moment. and the waitable latency of the incoming task After the task calculation is completed in space L, the task will be processed by the fog node or edge. The task offloading mode is based on the completion time. The decision-making process can be expressed by the following formula: Among them, 0 indicates local execution, 1 indicates that the task has timed out when it was offloaded, and is represented by . It represents the task that is about to be executed. i and j are both serial numbers, belonging to V (vehicle node set) and R (fog node resource set), respectively.

[0029] The model introduces the term "resource usage cost", which is a comprehensive calculation and definition of the three values ​​of vehicle movement mode, available resource status and the amount of historical task unloading data in "step 3.2". That is, movement, resources and historical data together affect the resource usage cost. Then, the "total cost of resource usage" is the total cost generated by the "vehicle movement mode, available resource status and the amount of historical task unloading data" of the task; the "partial task cost of the fog end" is the partial cost generated by the vehicle movement mode, available resource status and the amount of historical task unloading data at the fog end.

[0030] The following formula explains how to calculate resource usage costs using movement, resources, and historical data: According to the characteristics of fog nodes performing tasks in fog clusters, such as high mobility, unpredictable behavior, and limited resources, In the process of resource allocation, the stability of task execution is quantified by calculating the resource usage cost. Status information including movement information , the available computing resources of fog nodes The mobile information and computing resources can be expressed as: , Assume that the amount of historical task offloading data is ,Will Import resource usage cost calculation mode, task The total system resource usage cost calculation state space can be expressed as: .

[0031] The formula is defined as follows: in, is the total energy consumed by the fog node in the i-th period; Unloading factor, represents the remaining resources of the current vehicle at time t, Represents the resource threshold. When the remaining resources of the vehicle are lower than the threshold, it will no longer actively participate in the task offloading decision process and only retain its existing offloading mode until the remaining resources of the vehicle are higher than the threshold. It will participate in the task offloading decision process again.

[0032] This embodiment defines the available resource cost of the vehicle price 1, the fog end uses the resource cost alone Fog_pric Set e to n , and then the resource cost required for fog / edge task offloading can be calculated. The resource cost required for partial task offloading at the fog end is The remaining fog end processing tasks Some resource costs are: in, is the total unloading cost of the fog end, The remaining resource cost of the fog end processing task, for Allocate optimal blocks of computing resources.

[0033] Here is a definition of a resource block. A resource block can be defined as a division unit of resource usage cost. The resource usage cost can be composed of many equally divided resource blocks.

[0034] The following are the model details included in step 3.3: 1. Calculation of weighted cost of offloaded tasks: Calculation time of vehicle unloading task at the next moment The payment weight for resource usage is set to , when allocating resource blocks at a time, the maximum number of blocks that can be allocated shall not exceed, Allocate the best computing resource block for the current task. The weighted cost model is expressed as: in, is the weight of the delay, is the actual latency of fog terminal processing tasks, For the task The weighted cost of .

[0035] 2. Calculation of waiting delay of offloaded tasks: The allocated resource blocks must not exceed the maximum number of resource blocks that can be allocated. When the remaining resource blocks are not enough to meet the needs of the next task, it is necessary to wait for the previous task to be completed until it is satisfied before processing. When the resource requirements are met, the waiting delay is the maximum calculation delay of the previously completed task. The delay of subsequent tasks will also accumulate the previous waiting delay, so the waiting delay It is expressed as: in, is the cumulative delay of all tasks executed by the fog end. It is the latency of fog end processing tasks.

[0036] 3. Calculation of vehicle unloading task time: Combined with the vehicle movement pattern and available resource status, the time cost required for the fog end and the edge end to perform the vehicle unloading task at the next moment is calculated, and then the calculation time of the vehicle unloading task is obtained: where represents the efficiency of the i-th fog node, Represents the corresponding workload, where the workload is affected by the fog node resource cost capacity restrictions. represents the task arrival rate. Task t is processed by a fog node or an edge node in the computation space L with a probability of p. Indicates the task deadline, using Description, where , the actual processing time of task i must meet the deadline of this task.

[0037] 4. Vehicle task unloading determination: Task The best offloading decision is to perform calculation processing immediately after uploading. Waiting will reduce the task calculation processing time. The delay weight If the task waiting time is equal to or greater than the deadline, , ,so and The following relationship is satisfied: The artificial intelligence model algorithm is selected from the Long Short-Term Memory (LSTM) model. The LSTM network is a recursive neural network. Combined with the LSTM-based deep learning algorithm, it can realize the prediction of task-related parameters. LSTM is suitable for long-term data sequence modeling. The algorithm can effectively retain the long-term features in the data and is suitable for edge computing vehicle trajectory prediction. Assume that the amount of historical task data generated by the vehicle node is ,Will Import the LSTM prediction model for the task data volume at the next moment The goal of this prediction model is to improve the prediction accuracy of task information as much as possible, that is, .

[0038] Step 4 specifically includes the following steps: Step 4.1: When the task offloading decision is accepted, the edge vehicle offloads the task to the fog node, and all accepted tasks are saved in the local queue of the fog node until the resources meet the needs of stable task execution; Step 4.2: After the task is executed, all output results are stored in the result queue of the fog node; Step 4.3: The fog node’s scheduler sends the output results back to the edge vehicle using a first-come, first-served algorithm.

[0039] In this embodiment, the digital twin module integrates the physical entity, sensor communication, signal conditioning, data acquisition and control system by establishing the physical model and three-dimensional model of the fog node. After performing dynamic edge vehicle characteristics analysis, the data storage device of the fog node collects and stores different basic equipment and network data information.

[0040] Ultimately, the system can obtain four independent status data: vehicle status, fog node status, network status, and wireless channel status. The digital twin module will extract data in the form of mapping to calculate, define, and describe the basic characteristics of physical components and networks in the edge computing vehicle network. Utilizing the personalized adaptability of the fog terminal and the real-time control execution and perception capabilities of the edge terminal, the module uses the fog node as the hub of service requests and provides a highly available dynamic consistency model in the operation of the Internet of Vehicles.

[0041] The following is an explanation of the modeling of digital twin-assisted fog nodes and the methods and formulas for calculating resources within the model: The digital twin assisted environment generated by the digital twin module assisted by the fog node can be expressed as ,in is the edge vehicle node, represents the fog node set, and the mobile edge nodes are represented by express, represents the task queue at time t, Represents the length of the task queue. Assuming L represents the distance between adjacent fog nodes, the parameters closely related to the edge vehicle node v in the digital twin scenario are represented by the following set: (1) in, Indicates the queue status. represents the local computing resources of the vehicle, Indicates the current position of the vehicle. represents the task mode at time t, Represents other parameters related to the vehicle. The task queue in the digital twin mode can be expressed as The input data size of the task is Indicates. Completed the task The required processing capacity is expressed as (in CPU cycles). The maximum delay for this task is ,The task processing adopts the first-come-first-served mode. The available resources of all fog nodes are expressed as , R is the fog node resource set, r represents the fog node resource number, and the resources can be used for tasks The internal computing power provided is , for each available resource number n.

[0042] The specific model and parameter calculation method are as follows: 1. Vehicle edge computing communication model In step a, the basic model is as follows: Calculate the time consumption of data transmission between all connected edge vehicle nodes and fog nodes, and the wired or wireless data offloading rate between two nodes is .

[0043] (2) Among them, i and j are serial numbers, belonging to V (a collection of vehicle nodes) and R (Fog node resource collection), is the data offloading rate of the communication between the vehicle node and the fog node resources. The data rate between the edge vehicle node and the fog node resources is considered in a symmetric mode. Represents the data offloading rate between fog node resources.

[0044] Calculation of data offloading rate in wireless state based on Shannon-Hartley theorem : (3) in Indicates the transmit power, represents the spectral density of the disturbance level, Indicates the current channel capacity provided.

[0045] 2. Vehicle edge computing task model In step b, the process of perception task decision-making is as follows: for each edge vehicle node , a computing task is composed of a triple of MIPS (millions of instructions per sec) mode, input data (kbs) and task set S. Each task t can be expressed as , where s is the input computation size and c is the number of CPU cycles required to complete the operation.

[0046] 3. Vehicle edge computing energy consumption model In step b, the energy consumption model is mainly used to analyze the energy consumption in local computing and task offloading.

[0047] 1) Local computing mode: the amount of resources used by the i-th task for (4) in The power consumption rate for each CPU cycle is ; Available tasks for resources The internal computing power provided, The resources required for this subtask, is the natural logarithm, is the length of the task queue, and x is the sequence number of a subtask belonging to Q.

[0048] 2) Unloading calculation mode: The resource amount of the i-th resource is calculated as (5) in is the transmission power, is the power consumption rate of the CPU cycle, and its value is ; is the fog node set, is the execution cycle required for the task, CPU cycles for waiting task queue execution.

[0049] 4. Vehicle edge computing delay model In step b, according to the calculation strategy ,The tasks are executed locally using the onboard computing units, with additional resource provisioning through nearby fog nodes.,Therefore, the actual task offloading patterns can be divided into executing the tasks locally or partially,delegating them to nearby fog nodes.

[0050] 1) Local computing mode: computing time of the i-th task Calculated as (6) in, is the computing capacity of the specified resource, The number of CPU cycles required to complete the task, is the total number of waiting CPU cycles in the task queue executed locally at the edge vehicle node.

[0051] 2) Offloading computing mode: When the i-th task is delegated to a nearby fog node, the computing time The method is: (7) represents the data offloading rate between edge nodes and fog nodes, represents the computing energy consumption of allocating resources to fog nodes, is the task input size, is the execution cycle required for the task, CPU cycles for waiting task queue execution.

[0052] 5. Vehicle edge computing system load model In step b, the system load model is used to measure the demand for computing tasks and the current workload of the device, and two models, local computing and offload computing, are used for processing according to the load determination result. 1) Local computing mode: The local cost is: ,in and is the time and energy coefficient calculated for the weighted overhead, satisfying and .

[0053] 2) Offload calculation mode: The offload cost is ,in and is the time and energy coefficient calculated for the weighted overhead, satisfying and .

[0054] 6. Vehicle edge computing task offloading model In step c, the process of completing the task offloading decision is as follows: After completing the task calculation in space L, a decision is made whether the task is processed by the fog node or locally. The task offloading mode is based on the completion time. Make a decision. The decision-making process can be expressed by the following formula: (8) Among them, 0 means local execution, and 1 means that the task has timed out when being offloaded, which is represented by .

[0055] This embodiment adopts a task offloading mode with a prediction mechanism to guide the execution of task offloading decisions by predicting the vehicle's location and load. This mode helps the digital twin-assisted mobile edge computing system to perform efficient resource management.

[0056] In step d, when the system optimization goal is to minimize the system overhead, it is particularly important to ensure the workload of fog nodes while maintaining the node service quality and avoiding frequent task delivery. The task offloading decision can be expressed as: (9) in represents the efficiency of the i-th fog node, Indicates the corresponding workload, at this time the workload is limited by the fog node capacity restrictions. represents the task arrival rate. Task t is processed by fog nodes or local nodes in the computation space L with probability p. Indicates the task deadline, using Description, where , the actual processing time of task i must meet the deadline of this task.

[0057] Formula 9 belongs to step d and is used to calculate the fog node capacity based on the fog node capacity. ” and “Task Deadline ” Two aspects of constraints are imposed to execute the task offloading decision. Only when the constraint of Formula 9 is satisfied can the fog node execute step d.

[0058] The advantages of the solution provided in this embodiment are compared with various vehicle edge computing methods disclosed in the prior art, and the results are shown in the following table.

[0059] Table 1 Comparison summary of existing solutions and solutions of this embodiment Note: In the above table, the meaning of "IoT communication (resource accessibility)" is that each node in the IoT network (such as devices, sensors, servers, etc.) can send, receive and process data according to the predetermined protocols and mechanisms, ensuring that resources can be accessed and used in a timely and accurate manner when needed. Some existing technologies do not consider the issue of resource accessibility.

[0060] The literature corresponding to each existing method is as follows: Existing method one: Z. Wang, R. Gupta, K. Han, H. Wang, A. Ganlath, N. Ammar, and P. Tiwari, "Mobilitydigital twin: Concept, architecture, case study, and future challenges," IEEE Internet of Things Journal, 2022. Existing method two: T. Liu, L. Tang, W. Wang, Q. Chen, and X. Zeng, "Digital-twinassisted task offloadingbased on edge collaboration in the digital twinedge network," IEEE Internet of Things Journal, vol. 9, no. 2, pp. 1427-1444, 2021. Existing method three: AA Alahmadi, TE El-Gorashi, and JM Elmirghani, "Energy efficient and delay aware vehicular edge cloud," in 2020 22 nd International Conference on Transparent Optical Networks (ICTON). IEEE, 2020, pp. 1-4. Existing method four: P. Dai, K. Hu, X. Wu, H. Xing, and Z. Yu, “Asynchronous deep reinforcement learning for data-driven task offloading in VECempowered vehicular networks.” in INFOCOM, 2021, pp. 1-10. Existing Method 5: W. Zhan, C. Luo, J. Wang, C. Wang, G. Min, H. Duan, and Q. Zhu, “Deep-reinforcement-learning-based offloading scheduling for vehicular edge computing,” IEEE Internet of Things Journal, vol. 7, no. 6, pp. 5449–5465, 2020. Existing Method 6: H. Zhang, Y. Yang, X. Huang, C. Fang, and P. Zhang, “Ultra-low latency multi-task offloading in mobile edge computing,” IEEE Access, vol. 9, pp. 32 569–32 581, 2021. Existing Method 7: X. Xu, B. Shen, S. Ding, G. Srivastava, M. Bilal, M. R. Khosravi, V. G. Menon, M. A. Jan, and M. Wang, “Service offloading with deep q-network for digital twin-empowered internet of vehicles in edge computing,” IEEE Transactions on Industrial Informatics, vol. 18, no. 2, pp. 1414-1423, 2020. Existing Method 8: D. Van Huynh, V.-D. Nguyen, S. R. Khosravirad, V. Sharma, O. A. Dobre, H. Shin, and T. Q. Duong, “Urllc edge networks with joint optimal user association, task offloading and resource allocation: A digital twin approach,” IEEE Transactions on Communications, 2022. Existing method nine: J. Wang, J. Hu, G. Min, W. Zhan, Q. Ni, and N. Georgalas, “Computation offloading in multi-access edge computing using a deep sequential model based on reinforcement learning,” IEEECommunications Magazine, vol. 57, no. 5, pp. 64–69, 2019. Existing method ten: X. Yuan, J. Chen, N. Zhang, J. Ni, FR Yu, and VCLeung, “Digitaltwin-driven vehicular task offloading and irs configuration in the internet of vehicles,” IEEE Transactions on IntelligentTransportationSystems, 2022. As shown in the above table, different from other existing research results and solutions in this field, this solution is the first to handle task offloading requirements in a predictive mode in fog / edge mode, and is the first model that simultaneously has the three features of digital twin application, minimum latency optimization mode and resource accessibility.

[0061] In summary, the vehicle edge computing predictive task offloading method and system of the present invention combines the technologies of digital twin and fog union. Through the predictive task offloading method, it can reduce the task completion delay, improve the balance of fog node resource allocation, adapt to virtual and real errors and improve the VEC system calculation rate. Therefore, the present invention has a good application prospect.

Claims

1. A digital twin-assisted vehicle edge computing predictive task offloading method, characterized in that: Perform the following steps on the fog node: Step 1: Sense all task offloading requirements in the domain; Step 2: Receive the available computing resource status from the edge vehicle and forward it to the digital twin module; Step 3: Perform parameter simulation through the digital twin module, and judge whether the fog node accepts the task offloading demand based on the parameter simulation results and the actual resource inventory, and obtain the task offloading decision; Step 4: For the accepted task offloading decision, the edge vehicle offloads the task to the fog node for task execution, and then sends the output result back to the edge vehicle.

2. The digital twin-assisted vehicle edge computing predictive task offloading method according to claim 1 is characterized in that: In step 1, the fog node is responsible for maintaining a list of edge vehicles that access and leave the fog node's communication network range; when an edge vehicle accesses, it is initialized and sends its task offloading requirements to the fog node.

3. The digital twin-assisted vehicle edge computing predictive task offloading method according to claim 1 is characterized in that: In step 3, the digital twin module is configured to: create the network topology of all connected nodes and the task interaction relationship, use the digital twin model to perform parameter simulation on the task offloading decision and data transmission method of the actual fog node or edge node, and realize the prediction of the task computing capacity of the fog node.

4. The digital twin-assisted vehicle edge computing predictive task offloading method according to claim 1 is characterized in that: Step 3 specifically includes the following steps: Step 3.1: Input the historical task offloading data generated in the domain into the AI ​​model integrated in the digital twin model; Step 3.2: Calculate the computing time required to execute the vehicle task at the next moment by combining the vehicle movement pattern, the available resource status and the amount of historical task offloading data; Step 3.3: Compare the calculation time of the vehicle task at the next moment and the task waiting delay contained in the task offloading requirement. If the calculation time of the vehicle task at the next moment is lower than the task waiting delay, the fog node accepts the task offloading requirement, otherwise it rejects it.

5. The digital twin-assisted vehicle edge computing predictive task offloading method according to claim 4 is characterized in that: In step 3.2, the artificial intelligence model algorithm is selected from LSTM, RNN, CNN, Bayesian method or ARIMA time series model.

6. The digital twin-assisted vehicle edge computing predictive task offloading method according to claim 4 is characterized in that: In step 3.3, the criteria for judging whether the fog node accepts the task offloading decision also include: when the edge vehicle is directly connected to the fog node and the fog node is performing the task, the task offloading decision can be accepted, otherwise it is rejected.

7. The digital twin-assisted vehicle edge computing predictive task offloading method according to claim 1 is characterized in that: Step 4 specifically includes the following steps: Step 4.1: When the task offloading decision is accepted, the edge vehicle offloads the task to the fog node, and all accepted tasks are saved in the local queue of the fog node until the resources meet the needs of stable task execution; Step 4.2: After the task is executed, all output results are stored in the result queue of the fog node; Step 4.3: The fog node’s scheduler sends the output results back to the edge vehicle.

8. The digital twin-assisted vehicle edge computing predictive task offloading method according to claim 1 is characterized in that: The following steps are performed locally on the edge vehicle: Step a: The edge vehicle generates a task and stores it in a local queue; Step b: The resource manager of the edge vehicle perceives the decision basis of task offloading in real time, calculates the task requirements and the current workload of the device, and decides whether the task is processed locally or by the fog node. For the task processed by the fog node, the task offloading requirement is obtained; the decision basis of task offloading includes vehicle metadata, task data, fog resources and task deadline information; Step c: temporarily storing the task offloading demand in the offloading queue; Step d: When the edge vehicle is connected to the fog node, it sends the task offloading requirement.

9. A system for implementing the digital twin-assisted vehicle edge computing predictive task offloading method according to any one of claims 1 to 8, characterized in that: include: Fog nodes and at least one edge vehicle; The fog node is connected to a digital twin module.

10. A computer-readable storage medium, characterized in that: Stored thereon are: a computer program for implementing the digital twin assisted vehicle edge computing predictive task offloading method as described in any one of claims 1-8, or a computer program for implementing the system as described in claim 9.

Images