Cooperative sensing task unloading method based on vehicle-mounted edge computing

Abstract

The invention discloses a cooperative perception task unloading method based on vehicle-mounted edge calculation, and aims to improve the perception quality and calculation efficiency of an automatic driving system. The method optimizes a task unloading strategy through information value evaluation and redundancy perception suppression; the method comprises the following steps: firstly, predicting a vehicle motion track and estimating a region of interest (ROI) by using a Kalman filter, and calculating a VOI value of perceptual data so as to quantify the contribution degree of the data to a decision; secondly, redundant data are recognized by calculating the vehicle view overlapping degree, and transmission and calculation burdens of redundant sensing tasks are reduced; then, based on an integer linear programming (ILP) model, optimizing an unloading decision, maximizing perception benefits, and ensuring that tasks are distributed to an optimal computing node; a dynamic unloading scheduling strategy based on VOI is further adopted, and a task unloading path is dynamically adjusted according to the real-time evaluation result and the network condition; besides, a performance feedback and optimization mechanism is designed for the system, the unloading execution effect is continuously monitored, a strategy is adjusted according to historical data, and long-term stable operation is achieved.

Description

Technical Field

[0001] The present application belongs to the field of intelligent information processing and relates to a collaborative perception task offloading method based on vehicle-mounted edge computing. Background Art

[0002] With the rapid development of intelligent transportation systems (ITS) and autonomous driving technologies, vehicular edge computing (VEC), as an effective computing and communication architecture, has become an important means to improve the perception capability and response speed of autonomous driving systems. By offloading computing tasks to the edge nodes of the vehicle network, VEC can reduce the vehicle's dependence on traditional data centers, reduce latency, and improve the system's real-time processing capabilities. In the autonomous driving environment, cooperative perception (CP) is a crucial technology that allows different vehicles to jointly build a more comprehensive and accurate traffic environment model by sharing perception information, thereby significantly enhancing road safety, optimizing driving paths, and improving traffic flow. However, the current application of VEC in cooperative perception task offloading still faces many challenges. Traditional perception task offloading solutions usually focus on the number of tasks or load distribution, while ignoring the quality and information value of the perception task itself. With the continuous increase in the amount of perception data and sharing frequency between autonomous driving vehicles, the problems of perception redundancy and information overload have become more serious, greatly reducing the efficiency and reliability of the cooperative perception system.

[0003] Existing collaborative perception task offloading methods often rely on the balance between the number of tasks and the network load, but lack an accurate assessment of information quality when making task offloading decisions. In vehicle networks, due to the dynamics of vehicles and the complexity of the traffic environment, task offloading is not only limited by computing resources, but also affected by network performance factors such as transmission bandwidth and latency. Existing solutions fail to fully consider the heterogeneity and diversity of perception information from each vehicle, resulting in the transmission and processing of redundant data, increasing the computational burden of the system, and reducing the accuracy of real-time decision-making. More importantly, traditional offloading strategies ignore the important factor of information value (VOI). Information exchange between vehicles should not only be based on the quantity of information, but should give priority to the effectiveness of the information and its contribution to the current decision.

[0004] In order to effectively solve these problems, this application proposes a collaborative perception task offloading method based on vehicle edge computing (VEC), which combines information value evaluation with redundant perception suppression, uses a Kalman filter to predict vehicle motion trajectory, and optimizes offloading decisions through an integer linear programming (ILP) model. Summary of the invention

[0005] In order to overcome the above defects, this application proposes a collaborative perception task offloading method based on vehicle-mounted edge computing. The specific steps of this application are as follows:

[0006] S1, Value of Information (VOI) evaluation, uses Kalman filter to predict the vehicle's motion trajectory, estimates the region of interest (ROI), and evaluates the value of information (VOI) based on the data's contribution to the decision, providing a quality measure for subsequent task offloading decisions;

[0007] S2, perception task redundancy assessment, analyzes and identifies redundant data in perception tasks, reduces redundant perception by calculating vehicle field of view overlap, improves system efficiency and reduces computing and transmission burdens;

[0008] S3, task offloading strategy optimization, optimizes offloading decisions through the integer linear programming (ILP) model, maximizes the perceived benefit, and ensures that task offloading selects the optimal computing node based on the vehicle's VOI value and timeliness requirements;

[0009] S4, dynamic unloading scheduling based on information value, dynamically adjusts the unloading strategy according to real-time VOI evaluation, network conditions and vehicle location to ensure that the system maintains optimal unloading efficiency and perceived quality under changing environments;

[0010] S5, performance feedback and optimization, monitors the offloading execution effect, collects performance data, optimizes system parameters through feedback mechanism, and ensures the long-term stable and accurate operation of task offloading strategy.

[0011] The technical features and improvements of this application are:

[0012] For step S1, the present application first predicts the motion trajectory of the vehicle through a Kalman filter (KF) to estimate the future position and region of interest (ROI) of each vehicle. The Kalman filter uses the current state estimate and observation information to correct the position and speed of the vehicle through a recursive algorithm to improve the accuracy of the prediction. The vehicle's region of interest is calculated based on its predicted trajectory and potential obstacles, traffic facilities and other factors in the environment. For each vehicle, the system calculates the area that the vehicle may focus on in the future. The value of information (VOI) is a quantitative assessment of the quality of perception data, usually considering the impact of the data on the current decision. VOI can be calculated using the following formula:

[0013] VOI i =f(Δx i ,Δv i ,Δt) (1)

[0014] Among them, VOI irepresents the information value of vehicle i, Δx i is the error between the current position and the predicted position of vehicle i, Δv i is the speed prediction error of vehicle i, and Δt is the prediction time window. VOI not only considers the predicted position of the vehicle, but also the timeliness and accuracy of the data and its impact on the decision-making of surrounding vehicles. This step provides a basis for subsequent task offloading decisions, allowing the system to prioritize high VOI data.

[0015] For step S2, the present application identifies and reduces redundant perception by calculating the overlap of perception data between vehicles. Redundant perception refers to multiple vehicles sharing the same or highly overlapping information. This data does not add value in improving traffic perception accuracy, so it needs to be removed from the unloading task. In order to evaluate redundancy, the overlapping area of ​​vision between vehicles can be used for calculation. Assume that the perception field of view of vehicles i and j is F respectively. i and F j , the overlapping area F i∩j It can be obtained by geometric calculation. The overlap ρ is calculated by the following formula:

[0016]

[0017] Among them, ρ i,j represents the overlap between the perception fields of vehicles i and j, |F i ∩F j | represents the area of ​​overlapping visual fields, |F i ∪F j | represents the union area of ​​the two vehicles’ fields of view. By calculating the overlap of each pair of vehicles, the system can identify which perception tasks are redundant and prioritize them according to the overlap, thereby deciding whether to reduce the transmission or processing of redundant data and optimize computing resources.

[0018] For step S3, this application optimizes the task offloading strategy through an integer linear programming (ILP) model. Based on the VOI evaluation and timeliness requirements of each vehicle, the goal of this step is to maximize the cooperative perception benefits of each vehicle in the system and reduce invalid computing tasks and redundant data transmission. Assuming that there are N vehicles, the set of tasks that each vehicle i needs to offload is T i , the VOI value of each task is VOI i And there are M edge computing nodes j, each with a processing capacity of C j , the task offloading problem can be modeled as an ILP problem:

[0019]

[0020] where x ij For task T iWhether to offload to edge computing node j, if offloading, then x ij =1, otherwise x ij =0, VOI i For task T i The information value of C j is the computing power of node j.

[0021] This application is designed with the following constraints:

[0022] (1) Each task can only be assigned to one node:

[0023]

[0024] (2) The computing power of each node cannot exceed its maximum processing capacity:

[0025]

[0026] By solving this ILP problem, the system is able to select the optimal offloading node for each vehicle's perception task to maximize the information value and ensure the efficient use of computing resources.

[0027] For step S4, this application proposes dynamic offloading scheduling based on information value to dynamically schedule task offloading. Due to the dynamic movement of vehicles and changes in the traffic environment, the task offloading strategy needs to be adjusted according to the real-time VOI evaluation, network status, and the load of edge computing nodes. In this step, the system optimizes task offloading in real time based on the current information value (VOI) and network conditions. Set the current VOI of vehicle i and node j to VOI ij , and consider the current load and bandwidth B of the node j The goal is to maximize the total information value at the current moment and ensure load balancing of each node. This process can be modeled by the following optimization formula:

[0028]

[0029] Among them, y ij Indicates whether the task of vehicle i is offloaded to node j. If unloaded, y ij =1, otherwise y ij =0, VOI ij Represents the real-time information value of vehicle i at node j.

[0030] This application is designed with the following constraints:

[0031] (1) Node load constraints:

[0032]

[0033] (2) Each task must be offloaded to a certain node:

[0034]

[0035] By dynamically adjusting the offloading strategy, the system can respond to changes in vehicle movement and fluctuations in the network environment in real time, ensuring optimal task offloading scheduling at every moment.

[0036] The collaborative perception task offloading method based on vehicle edge computing (VEC) of the present application solves the problem of neglecting redundant data and insufficient evaluation of information value in the perception task offloading process in the prior art, and has the following advantages:

[0037] (1) The collaborative perception task offloading method based on vehicle edge computing (VEC) proposed in this application effectively reduces the transmission and processing of redundant data by giving priority to the value of information (VOI) of the perception task rather than simply relying on the number of perception tasks. By using the Kalman filter for future position prediction and area of ​​interest estimation, the system can accurately evaluate the contribution of each vehicle's perception data to traffic decision-making, thereby maximizing the overall perception quality of the system and significantly improving the benefits of collaborative perception;

[0038] (2) This application introduces a dynamic unloading scheduling mechanism based on information value, which can optimize the task unloading path in real time according to the vehicle's motion state, network conditions and edge computing node load. This solution ensures efficient use of computing resources through load balancing and bandwidth scheduling, and effectively avoids the problem of computing node overload or insufficient bandwidth, thereby improving the overall computing and transmission efficiency of the system. ;

[0039] (3) This application combines performance feedback mechanism and historical data learning. Through continuous monitoring and optimization, it can adjust the offloading strategy and system parameters in real time to ensure the long-term stable operation of the system in a complex and dynamic traffic and network environment. The feedback mechanism adjusts the offloading decision based on real-time performance data, so that the system can adapt to changing network conditions, traffic conditions and computing requirements, ensuring that the task offloading strategy remains accurate, stable and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flow chart of the vehicle edge computing system architecture in this application.

[0041] Figure 2 This is a diagram of the information value assessment process in this application.

[0042] Figure 3 Evaluation and optimization diagram for the redundancy perception task in this application.

[0043] Figure 4 This is a diagram of the dynamic unloading scheduling process in this application. DETAILED DESCRIPTION

[0044] The present application is further described in detail below with reference to the accompanying drawings and specific implementation methods:

[0045] A collaborative sensing task offloading method based on vehicle-mounted edge computing, such as Figure 1 As shown, it is a flowchart of collaborative perception task offloading based on vehicle-mounted edge computing of this application, and the method includes:

[0046] S1, accurately predict the motion trajectory of each vehicle in order to estimate its future motion position and its perceived region of interest (ROI). To this end, this application uses a Kalman filter (KF) to estimate and update the state of the vehicle. The Kalman filter is a recursive estimation algorithm that uses the current state estimate and sensor measurement data to predict the future motion trajectory of the vehicle. Specifically, the vehicle's state vector x k It can be expressed as:

[0047]

[0048] Among them, p k is the position of the vehicle at time k, v k is the speed. When performing state estimation, the Kalman filter uses a dynamic model to predict the state of the vehicle at the next moment:

[0049] x k+1 =F·x k +B·u k +w k (10)

[0050] Among them, F is the state transfer matrix, B is the control input matrix, and u k is the control vector, w k is process noise. The Kalman filter continuously iteratively updates the vehicle state to improve the accuracy of future position predictions. Then, based on the predicted trajectory and current environmental information, the system estimates the vehicle's region of interest ROI i , that is, the area that the vehicle may focus on in the future. The vehicle's area of ​​interest is dynamically updated by analyzing the vehicle's trajectory, speed, and relative position to other traffic objects (such as other vehicles, obstacles, etc.). On this basis, the calculation of the value of information (VOI) can provide a basis for unloading decisions. The value of information is a quantitative assessment of the quality of collaborative sensing data, usually calculated based on the accuracy and timeliness of the sensing data and its contribution to traffic decisions. Assume that for each vehicle i, its value of information VOI i It can be expressed as:

[0051] VOIi = f(Δx i , Δv i , Δt) (11)

[0052] where Δx i is the position error of vehicle i, i.e., the difference between the predicted position and the actual position; Δv i is the speed prediction error of vehicle iii; and Δt is the length of the prediction period. The evaluation of the information value reflects the actual contribution of the perception data obtained by each vehicle to the perception of the surrounding environment, providing a quantitative indicator for subsequent offloading decisions.

[0053] S2. Identify and reduce redundant information in the perception tasks between vehicles to optimize the utilization of data transmission and computing resources. Redundant perception refers to the same or highly overlapping perception data shared among multiple vehicles, which has no additional contribution to improving the perception effect. To calculate the redundancy of the perception tasks, we first analyze the perception field of view areas F i and F j of each vehicle, which represent the range of the traffic environment that vehicles i and j can observe at a given moment. The overlapping area F i∩j between the two vehicles can be obtained through geometric calculation, representing the part they perceive in common. The overlap degree ρ i,j is calculated by the following formula:

[0054]

[0055] where |F i ∩F j | represents the area of the overlapping field of view, and |F i ∪F j | represents the union area of the fields of view of the two vehicles. When ρ i,j is large, it indicates that there is a high redundancy in the perception tasks between these two vehicles, and the shared perception data can be considered for reduction. According to the magnitude of the redundancy, the system can dynamically adjust the task offloading strategy to reduce the transmission of redundant perception tasks and improve the computing efficiency of the system.

[0056] S3. Construct an integer linear programming (ILP) model to optimize the task offloading decision. The offloading of vehicle perception tasks needs to consider multiple factors, including the information value of interest (VOI) of each vehicle, the timeliness requirements of the tasks, the network bandwidth, and the computing capabilities of the edge computing nodes. Assume that there are N vehicles in the system, the task set of each vehicle is T i , the VOI of each task is VOI i , and there are edge computing nodes, and the computing capabilities of each node are C j . The task offloading problem can be modeled as the following ILP optimization problem:

[0057]

[0058] where x ij For task T i Whether to offload to edge computing node j, if offloading, then x ij =1, otherwise x ij =0, VOI i For task T i The information value of C j is the computing power of node j. This optimization problem needs to meet the following constraints:

[0059] (1) Each task can only be assigned to one node:

[0060]

[0061] (2) The computing power of each node cannot exceed its maximum processing capacity:

[0062]

[0063] where load i It is task T i The computational load, C j is the computing power of node j. By solving this ILP problem, the system can select the optimal offloading node for each task, maximize the perceived benefit of task offloading, and ensure the reasonable allocation of computing resources.

[0064] S4 proposes dynamic offloading scheduling based on real-time value of information (VOI) and network conditions. The movement of vehicles, traffic conditions, and changes in the network environment require that the offloading strategy must be highly dynamic. In order to cope with these changes, this application designs a dynamic adjustment mechanism that adjusts the task offloading strategy in real time at each moment based on the vehicle's movement trajectory, network bandwidth, node load, and current VOI evaluation results.

[0065] Set the real-time information value of each task as VOI ij , represents the task T of vehicle i i The real-time information value on computing node j. In addition, considering the bandwidth limitation and computing load of each node, the goal of the system is to maximize the total information value after task offloading and ensure that the load of the node does not exceed its maximum processing capacity. The optimization objective function is:

[0066]

[0067] Among them, y ij Indicates whether the task of vehicle i is offloaded to node j. If unloaded, y ij =1, otherwise y ij=0, VOI ij Represents the real-time information value of vehicle i at node j.

[0068] This application is designed with the following constraints:

[0069] (3) Node load constraints:

[0070]

[0071] (4) Each task must be offloaded to a certain node:

[0072]

[0073] By adjusting the unloading schedule in real time, the system can cope with the dynamically changing traffic environment and network conditions, ensuring the optimality and real-time performance of task unloading.

[0074] S5, introduces a performance feedback mechanism to comprehensively monitor the execution effect of task offloading, and optimizes system parameters based on the performance data collected in real time, thereby ensuring the long-term stability and accurate operation of the task offloading strategy. During the task offloading process, the system continuously tracks and records multiple key performance indicators, including task offloading latency, edge computing node load status, bandwidth utilization, and information value (VOI). These indicators are the core data for measuring the execution effect of offloading tasks, which can reveal the bottlenecks and potential problems of the system in actual operation and help the system to make timely adjustments and optimizations. Task offloading latency D ij The calculation formula is:

[0075]

[0076] Among them, Load ij Represents task T i The computational load on node j, B j is the bandwidth of the node, TransmissionDelay ij is the transmission delay of the task from the vehicle to the node. This performance data provides the system with real-time feedback on task offloading. In addition, the system also calculates the bandwidth utilization rate. j Monitor to ensure efficient use of network resources. The calculation formula is:

[0077]

[0078] By collecting these performance data, the system can understand the operation status of the offload task in real time, identify possible problems such as excessive load, high latency or insufficient bandwidth utilization, and provide a basis for subsequent optimization.

[0079] Based on the performance data collected in real time, the system automatically adjusts the offloading strategy and system parameters through a feedback mechanism to improve overall performance and resource utilization. When the load on a certain node is monitored to be too high, the system can automatically offload the task to the node with a lighter load to achieve load balancing. In the case of long task offloading delays, the system will analyze the reasons, which may be bandwidth bottlenecks or insufficient computing resources, and adjust the bandwidth allocation for task offloading or select nodes with stronger computing power accordingly. The feedback mechanism makes decision adjustments through the following optimization formula:

[0080]

[0081] in, It is a task offloading decision after feedback optimization, reflecting the process of the system dynamically adjusting the task offloading path according to the current performance indicators. This adjustment process can ensure that the offloading strategy always maintains the optimal state under dynamically changing network and traffic conditions. At the same time, the system performs adaptive learning based on historical data to form a long-term performance optimization model. By accumulating and analyzing feedback data collected during long-term operation, the system continuously updates the optimization strategy and forms a historical behavior model to predict the effect of future task offloading and adjust the strategy in advance. The combination of historical data learning and feedback optimization can ensure the stable operation of the system in a complex and dynamic environment and adapt to the changing network conditions and computing needs. Through methods such as reinforcement learning, the system can autonomously identify and adjust the optimal offloading strategy to meet the needs of different scenarios.

[0082] Through multi-dimensional optimization, the system not only focuses on minimizing the task offloading latency, but also takes into account bandwidth utilization, node load balancing, and maximizing information value, thereby achieving global optimization of the task offloading process. The optimization of the offloading strategy is a comprehensive process that requires the system to balance multiple goals to ensure that each dimension is effectively improved. Ultimately, the system can maintain long-term stability and efficient operation through performance feedback and optimization mechanisms, ensuring that the task offloading strategy is always accurate and effective in various dynamically changing environments.

[0083] In summary, this application proposes a collaborative perception task offloading method based on vehicle-mounted edge computing, which effectively reduces redundant data, optimizes computing resource utilization, and ensures the long-term stability and efficient operation of the system in complex environments by prioritizing information value and dynamically optimizing offloading strategies.

[0084] Although the content of the present application has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present application. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present application can be made. Therefore, the protection scope of the present application should be limited by the appended claims.

Claims

1. A collaborative sensing task offloading method based on vehicle-mounted edge computing, its characteristics and Specific steps: S1, Value of Information (VOI) evaluation, uses Kalman filter to predict the vehicle's motion trajectory, estimates the region of interest (ROI), and evaluates the value of information (VOI) based on the data's contribution to the decision, providing a quality measure for subsequent task offloading decisions; S2, perception task redundancy assessment, analyzes and identifies redundant data in perception tasks, reduces redundant perception by calculating vehicle field of view overlap, improves system efficiency and reduces computing and transmission burdens; S3, task offloading strategy optimization, optimizes offloading decisions through the integer linear programming (ILP) model, maximizes the perceived benefit, and ensures that task offloading selects the optimal computing node based on the vehicle's VOI value and timeliness requirements; S4, dynamic unloading scheduling based on information value, dynamically adjusts the unloading strategy according to real-time VOI evaluation, network conditions and vehicle location to ensure that the system maintains optimal unloading efficiency and perceived quality under changing environments; S5, performance feedback and optimization, monitors the offloading execution effect, collects performance data, optimizes system parameters through feedback mechanism, and ensures the long-term stable and accurate operation of task offloading strategy.

2. According to claim 1, a collaborative sensing task offloading method based on vehicle-mounted edge computing is characterized in that: For step S1, the present invention first predicts the motion trajectory of the vehicle through a Kalman filter (KF) to estimate the future position and region of interest (ROI) of each vehicle; the Kalman filter uses the current state estimation and observation information to correct the position and speed of the vehicle through a recursive algorithm to improve the prediction accuracy. The vehicle's region of interest is calculated based on its predicted trajectory and potential obstacles and traffic facilities in the environment; for each vehicle, the system calculates the area that the vehicle may focus on in the future. The value of information (VOI) is a quantitative assessment of the quality of the perception data, usually considering the impact of the data on the current decision. The VOI can be calculated by the following formula: VOI i =f(Δx i ,Δv i ,Δt) (1) Among them, VOI i represents the information value of vehicle i, Δx i is the error between the current position and the predicted position of vehicle i, Δv i is the speed prediction error of vehicle i, Δt is the prediction time window, and VOI not only considers the predicted position of the vehicle, but also the timeliness and accuracy of the data and the impact on the decision-making of surrounding vehicles. This step provides a basis for subsequent task offloading decisions, allowing the system to prioritize high VOI data.

3. The method for collaborative sensing task offloading based on vehicle-mounted edge computing according to claim 1 is characterized in that: For step S2, the present invention identifies and reduces redundant perception by calculating the overlap of perception data between vehicles; redundant perception refers to multiple vehicles sharing the same or highly overlapping information, which has no added value in improving traffic perception accuracy, and therefore needs to be removed from the unloading task; in order to evaluate the redundancy, the overlapping area of ​​vision between vehicles can be used for calculation; the perception field of view of vehicles i and j is set to be F respectively. i and F j , the overlapping area F i∩j It can be obtained through geometric calculation, and the overlap ρ is calculated by the following formula: Among them, ρ i,j represents the overlap between the perception fields of vehicles i and j, |F i ∩F j | represents the area of ​​overlapping visual fields, |F i ∪F j | represents the union area of ​​the two vehicles’ fields of view. By calculating the overlap of each pair of vehicles, the system can identify which perception tasks are redundant and prioritize them according to the overlap, thereby deciding whether to reduce the transmission or processing of redundant data and optimize computing resources.

4. The method for collaborative sensing task offloading based on vehicle-mounted edge computing according to claim 1 is characterized in that: For step S3, the present invention optimizes the task offloading strategy through an integer linear programming (ILP) model; based on the VOI evaluation and timeliness requirements of each vehicle, the goal of this step is to maximize the cooperative perception benefits of each vehicle in the system and reduce invalid computing tasks and redundant data transmission; assuming that there are N vehicles, the set of tasks that each vehicle i needs to offload is T i , the VOI value of each task is VOI i And there are M edge computing nodes j, each with a processing capacity of C j , the task offloading problem can be modeled as an ILP problem: where x ij For task T i Whether to offload to edge computing node j, if offloading, then x ij =1, otherwise x ij =0, VOI i For task T i The information value of C j is the computing power of node j. This application is designed with the following constraints: (1) Each task can only be assigned to one node: (2) The computing power of each node cannot exceed its maximum processing capacity: By solving this ILP problem, the system is able to select the optimal offloading node for each vehicle's perception task to maximize the information value and ensure the efficient use of computing resources.

5. The method for collaborative sensing task offloading based on vehicle-mounted edge computing according to claim 1 is characterized in that: For step S4, the present invention proposes dynamic offloading scheduling based on information value to dynamically schedule task offloading; due to the dynamic movement of vehicles and changes in the traffic environment, the task offloading strategy needs to be adjusted according to the real-time VOI evaluation, network status, and load conditions of edge computing nodes. In this step, the system optimizes task offloading in real time based on the current information value (VOI) and network conditions; Set the current VOI of vehicle i and node j to VOI ij , and consider the current load and bandwidth B of the node j ; The goal is to maximize the total information value at the current moment and ensure the load balance of each node. This process can be modeled by the following optimization formula: Among them, y ij Indicates whether the task of vehicle i is offloaded to node j. If unloaded, y ij =1, otherwise y ij =0, VOI ij Represents the real-time information value of vehicle i at node j. This application is designed with the following constraints: (1) Node load constraints: (2) Each task must be offloaded to a node: By dynamically adjusting the offloading strategy, the system can respond to changes in vehicle movement and fluctuations in the network environment in real time, ensuring optimal task offloading scheduling at every moment.

Images