Task unloading method and system based on regional RSU in Internet of Vehicles environment

Abstract

The invention provides a regional RSU-based task unloading method and system in an Internet of Vehicles environment, and relates to the technical field of Internet of Vehicles, and the method specifically comprises the steps: constructing a vehicle trajectory prediction model, and generating a moving trajectory route map of a vehicle through the model; according to the moving track route map of the vehicle, the RSU closest to the vehicle at present is selected as an optimal unloading node, the distance between the vehicle and the RSU, the communication quality index of the vehicle and the RSU and the computing resource index of the RSU are obtained, and the income evaluation index of the RSU is generated; whether the unloading requirement of the RSU is met or not is judged according to the income evaluation index, when the unloading requirement of the RSU is not met, the RSU carries out unloading refusing operation on the user vehicle, the other RSUs are sequentially selected through a queuing algorithm to judge whether the unloading requirement is met or not till unloading can be carried out, and Nash equilibrium of the user vehicle and the RSU is achieved. And the problem of excessive time and energy consumption is solved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle networking, and specifically provides a task offloading method and system based on regional RSU in a vehicle networking environment. Background Art

[0002] In modern urban traffic systems, the rapid development of vehicle networking technology enables vehicles to interact with roadside units (RSUs) in real time, enhancing the intelligence of traffic management and services. Various data tasks generated during vehicle driving, such as navigation, environmental perception, and safety alerts, need to be offloaded to RSUs for calculation and processing in a timely manner. However, existing task offloading methods face numerous challenges. First, traditional offloading strategies often rely on static routing selection and fail to fully consider the dynamic movement characteristics of vehicles and real-time traffic conditions. This results in vehicles being unable to quickly find a suitable RSU for task offloading in high-density traffic or rapidly changing scenarios, leading to task delays or failures, and further affecting the vehicle's real-time decision-making ability.

[0003] In addition, there are also obvious deficiencies in evaluating the availability and performance of RSUs in existing technologies. Many systems fail to effectively consider various factors such as the communication quality, distance between vehicles and RSUs, and the computing resources of RSUs, resulting in inaccurate task offloading decisions. At the same time, when an RSU rejects a task offloading request, existing processing mechanisms often lack flexibility and fail to provide effective strategies for selecting alternative RSUs, causing unnecessary time waste during vehicle waiting. More importantly, advanced technologies such as queuing algorithms are not reasonably utilized to dynamically adjust offloading strategies, further exacerbating the inefficiency of the system. Therefore, there are obvious bottlenecks in the success rate and timeliness of existing task offloading technologies, which urgently need to be improved and optimized.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide a task offloading method and system based on regional RSU in a vehicle networking environment to solve the problems raised in the above background art.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A task offloading method and system based on regional RSU in a vehicle networking environment, and the specific steps include:

[0008] Step 1: Collect the historical trajectory map of the vehicle, construct a vehicle trajectory prediction model based on a convolutional neural network and a path planning algorithm, calculate the optimal path according to the starting position and the target position of the vehicle, and use the optimal path as the moving trajectory roadmap of the vehicle;

[0009] Step 2: According to the moving trajectory roadmap of the vehicle, select the RSU closest to the vehicle currently as the best offloading node, obtain the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, and after dimensionless processing of the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, generate the benefit evaluation index of the RSU;

[0010] Step 3: Judge whether it meets the offloading requirements of the RSU according to the benefit evaluation index. When the offloading requirements of the RSU are not met, the RSU performs a rejection offloading operation on the user vehicle, and uses a queuing algorithm to sequentially select the remaining RSUs to judge whether they meet the offloading requirements until it can be offloaded, realizing the Nash equilibrium between the user vehicle and the RSU;

[0011] Step 4: After the RSU finishes processing, return the calculation result to the user vehicle, and calculate the total energy consumed by this task offloading.

[0012] Further, the vehicle trajectory prediction model is constructed based on a convolutional neural network and a path planning algorithm. The construction process of the vehicle trajectory prediction model is specifically as follows:

[0013] Collect the historical trajectory map of the vehicle, and use it as input data. Use the trained convolutional neural network to extract features from the preprocessed historical trajectory map, construct the topological structure of the road network according to the extracted features, select the Dijkstra path planning algorithm, calculate the optimal path according to the starting position and the target position of the vehicle, and use the optimal path as the moving trajectory roadmap of the vehicle.

[0014] Further, to obtain the communication quality index between the vehicle and the RSU, the specific logic is as follows: within time T, use the wireless communication module of the vehicle to collect the signal strength during the communication between the vehicle and the RSU multiple times, and calculate the average value as the signal strength during the communication between the vehicle and the RSU, denoted as RSSI;

[0015] The user vehicle sends test data packets to the RSU, obtains the number of disconnections and reconnections during the data transmission process within time T and the corresponding duration of disconnection and reconnection. Mark the duration corresponding to the p-th disconnection and reconnection within time T as Tp, then the total duration N of disconnection and reconnection during the data transmission process within time T k The calculation expression is:

[0016]

[0017] Among them, p represents the numbering of the number of disconnections and reconnections during the data transmission process within time T, p = 1, 2, 3, 4, ……, q, and q is the total number of disconnections and reconnections within time T;

[0018] For N k and RSSI are dimensionless processed, and the communication quality index between the vehicle and the RSU is calculated according to the following formula:

[0019]

[0020] Among them, A is the communication quality index between the vehicle and the RSU, N k is the total duration of disconnection and reconnection during the data transmission process, and RSSI is the signal strength.

[0021] Furthermore, to obtain the computing resource index of the RSU, the specific logic is as follows: By referring to the hardware specification book and technical documents provided by the manufacturer, obtain the memory size and continuous power supply capacity of the RSU, record the memory size as G, and record the continuous power supply capacity as W.

[0022] After dimensionless processing the memory size and continuous power supply capacity, calculate the computing resource index of this RSU according to the following formula:

[0023] B = (α * G + β * W) 2

[0024] Among them, B is the computing resource index of the RSU, G is the memory size, W is the continuous power supply capacity, α and β are the proportionality coefficients of the memory size and continuous power supply capacity respectively, β > α > 0, and satisfy: α + β = 1.

[0025] Furthermore, according to the moving trajectory roadmap of the vehicle, select the RSU closest to the vehicle currently as the best offloading node, and collect the distance between the user vehicle and this RSU, denoted as X;

[0026] After dimensionless processing the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resources of the RSU, generate the revenue evaluation index of this RSU;

[0027] The formula for calculating the revenue evaluation index is as follows:

[0028]

[0029] Among them, QS is the revenue evaluation index, A is the communication quality index between the vehicle and the RSU, B is the computing resource index of the RSU, and X is the distance between the user vehicle and this RSU.

[0030] Further, compare the revenue evaluation index with a preset evaluation threshold:

[0031] If QS n ≥QY, it is determined that the current revenue evaluation index meets the offloading requirements of the RSU n , and the user vehicle selects the RSU n to perform task offloading. After the task processing is completed, the RSU n returns the calculation result to the user vehicle and calculates the total energy consumed for this task offloading;

[0032] If QS n <QY, it is determined that the current revenue evaluation index does not meet the offloading requirements of the RSU n , and the RSU performs a reject offloading operation on the user vehicle and uses the queuing algorithm to sequentially select the remaining RSUs to determine whether they meet the offloading requirements until offloading is possible;

[0033] In the formula, QS n is the revenue evaluation index of the RSU n , n is the index of the RSU, and QY is the preset evaluation threshold.

[0034] Further, the queuing algorithm is specifically: Define the user vehicle as V m , where m is the index of the user vehicle, define the task data volume generated in each time slot as U m (t), and the task is temporarily stored locally in the user vehicle and offloaded to the RSU for calculation and processing after the task is issued. The following data queue formula is formed locally in the user vehicle:

[0035] Q m (t + 1) = max{Q m (t) - L m (t) + U m (t), 0}

[0036] Among them, Q m (t + 1) represents the data volume existing on the user vehicle V m in the (t + 1)-th time slot, L m (t) represents the throughput of the user vehicle V m in the τ-th time slot, that is, the data volume offloaded to the RSU, and U m (t) represents the newly generated data volume of the user vehicle V m in the t-th time slot;

[0037] Define the communication sub-channel bandwidth between the user vehicle V m and the roadside unit RSU n as B m,n , when the user vehicle V mSelect the Road Side Unit (RSU) in the t-th time slot n When performing task offloading, calculate the amount of tasks m offloaded by the user vehicle V n to the RSU according to the following formula:

[0038]

[0039] where z m,n,t represents the amount of tasks m offloaded by the user vehicle V n to the RSU, τ is the fixed transmission time, P is the transmission power, G m,n,t is the channel gain of the user vehicle V m to the RSU n in the t-th time slot, and σ 2 is the noise power;

[0040] The formula for calculating the throughput of the user vehicle V m is as follows:

[0041]

[0042] where L m (t) represents the throughput of the user vehicle V m in the t-th time slot, X m,n,t represents the selection situation of the user vehicle V m for the RSU n in the t-th time slot. When X m,n,t = 1, it means that the user vehicle V m selects the RSU n to perform task offloading in the t-th time slot. If the RSU n rejects the task offloading, mark the task, and at the same time prohibit any more such tasks from being offloaded to the roadside unit RSU n , and use the queuing algorithm to transfer the task to the next RSU for offloading until the offloading condition is met.

[0043] Furthermore, after the RSU finishes processing, return the calculation result to the user vehicle and calculate the total energy consumed by this task offloading:

[0044] The formula for calculating the total energy consumed by task offloading is as follows:

[0045] The total energy consumed when the task is first offloaded to the RSU is:

[0046]

[0047] If the task is rejected by the RSU during the first offloading, the total energy consumed when selecting the y-th RSU for offloading according to the queuing algorithm is as follows:

[0048]

[0049] where τ is the fixed transmission time, P is the transmission power, and t R is the time consumed by the RSU to process the task, and t y represents the time required for data transmission from the user vehicle to the RSU when selecting the y-th RSU for task offloading. y is the index of the RSU, r represents the index of the RSU finally selected after multiple attempts when the user vehicle's first attempt to offload the task to the RSU is rejected, and ω represents the average task arrival rate, that is, the number of tasks sent by the user vehicle to the RSU per unit time.

[0050] The present invention also provides a task offloading system based on regional RSUs in a vehicle-to-everything environment. The task offloading system based on regional RSUs in a vehicle-to-everything environment is used to execute the above-mentioned task offloading method based on regional RSUs in a vehicle-to-everything environment, including:

[0051] A vehicle trajectory prediction module, which is used to collect the historical trajectory map of the vehicle, construct a vehicle trajectory prediction model based on a convolutional neural network and a path planning algorithm, calculate the optimal path according to the starting position and target position of the vehicle, and use the optimal path as the moving trajectory roadmap of the vehicle;

[0052] An RSU selection module, which is used to select the RSU closest to the vehicle currently as the best offloading node according to the moving trajectory roadmap of the vehicle, obtain the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, and generate the benefit evaluation index of the RSU after dimensionless processing of the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU.

[0053] An offloading requirement judgment module, which is used to judge whether it meets the offloading requirements of the RSU according to the benefit evaluation index. When the offloading requirements of the RSU are not met, the RSU performs a reject offloading operation on the user vehicle, and uses the queuing algorithm to sequentially select the remaining RSUs to judge whether they meet the offloading requirements until it can be offloaded, so as to achieve the Nash equilibrium between the user vehicle and the RSU.

[0054] A result return and energy consumption calculation module, which is used to return the calculation result to the user vehicle after the RSU finishes processing, and calculate the total energy consumed by this task offloading.

[0055] Compared with the prior art, the beneficial effects of the present invention are:

[0056] In this solution, the introduction of the vehicle trajectory prediction module enables the system to predict the vehicle's movement trajectory in advance, optimize the selection of RSU, and reduce task delays caused by poor distance and communication quality. Secondly, through the dimensionless benefit evaluation index, the RSU can more accurately determine whether the offloading requirements are met, thus achieving more efficient resource allocation and utilization. Finally, the use of the queuing algorithm ensures that when the offloading request is rejected, the vehicle task can be quickly transferred to other available RSUs, forming a non-cooperative game between the user vehicle and the RSU, and solving the problem of excessive time and energy consumption. Brief Description of the Drawings

[0057] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0058] Figure 2 It is a schematic diagram of the system module of the present invention;

[0059] Figure 3 It is a schematic diagram of the maximum energy consumption of the present invention compared with the traditional NOMA and full DRL task offloading methods within a certain period of time;

[0060] Figure 4 It is a schematic diagram of the maximum time delay of the present invention compared with the traditional NOMA and full DRL task offloading methods within a certain period of time. Detailed Embodiment

[0061] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to specific embodiments.

[0062] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meaning understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second", and similar terms used in the present invention do not indicate any order, quantity, or importance, but are only used to distinguish different components. The terms such as "including" or "comprising" mean that the elements or objects appearing before this word cover the elements or objects listed after this word and their equivalents, without excluding other elements or objects. The terms such as "connected" or "linked" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms such as "up", "down", "left", "right", etc. are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0063] Embodiment:

[0064] Please refer to Figure 1 and Figures 3-4, the present invention provides a technical solution:

[0065] A task offloading method based on regional RSU in the vehicle networking environment, the specific steps include:

[0066] Step 1: Collect the historical trajectory maps of vehicles, construct a vehicle trajectory prediction model based on a convolutional neural network and a path planning algorithm, calculate the optimal path according to the starting position and the target position of the vehicle, and use the optimal path as the moving trajectory roadmap of the vehicle;

[0067] In this embodiment, the vehicle trajectory prediction model is constructed based on a convolutional neural network and a path planning algorithm. The construction process of the vehicle trajectory prediction model is as follows:

[0068] Collect the historical trajectory maps of vehicles, and use them as input data. Use the trained convolutional neural network to extract features from the preprocessed historical trajectory maps, construct the topological structure of the road network according to the extracted features, select the Dijkstra path planning algorithm, calculate the optimal path according to the starting position and the target position of the vehicle, and use the optimal path as the moving trajectory roadmap of the vehicle.

[0069] In Step 1, by collecting the historical trajectory maps of vehicles and constructing a vehicle trajectory prediction model, the deep learning technology is used to accurately model the moving behavior of vehicles, significantly improving the understanding and prediction ability of the dynamic moving laws of vehicles. The advantage of this method is that the model trained based on historical data can capture the driving patterns of vehicles in complex traffic environments, thus providing high accuracy for the prediction of the next position. Compared with the prior art, it overcomes the limitations of traditional static models that cannot cope with dynamic traffic conditions and real-time changes, enabling the system to more intelligently select the optimal offloading path and node for vehicles.

[0070] In this solution, the implementation of Step 1 provides a data-driven decision-making basis for the overall task offloading process. By accurately predicting the moving trajectory of the vehicle, the system can effectively select the RSU closest to the vehicle and with the optimal communication quality for task offloading. This not only improves the offloading success rate, reduces the latency, but also promotes the resource utilization efficiency of the entire system. More importantly, the good prediction ability lays a foundation for the subsequent steps, enabling the vehicle to flexibly cope with various changes in the dynamic environment and improving the overall performance and reliability of the vehicle networking system.

[0071] Step 2: According to the moving trajectory roadmap of the vehicle, select the RSU closest to the vehicle currently as the best offloading node, obtain the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, and after dimensionless processing of the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, generate the benefit evaluation index of this RSU;

[0072] In this embodiment, the specific logic for obtaining the communication quality index between the vehicle and the RSU is as follows: within time T, the wireless communication module of the vehicle is used to collect the signal strength during the communication between the vehicle and the RSU multiple times, and the average value is obtained as the signal strength during the communication between the vehicle and the RSU, denoted as RSSI;

[0073] The user vehicle sends test data packets to the RSU, obtains the number of disconnections and reconnections during the data transmission process within time T and the corresponding durations of disconnections and reconnections. The duration corresponding to the p-th disconnection and reconnection within time T is calibrated as Tp, then the total duration N of disconnections and reconnections during the data transmission process within time T k The calculation expression is:

[0074]

[0075] where p represents the numbering of the disconnections and reconnections during the data transmission process within time T, p = 1, 2, 3, 4,..., q, q is the total number of disconnections and reconnections within time T, and the length of T is 10 seconds;

[0076] For N k and RSSI are dimensionless processed, and the communication quality index between the vehicle and the RSU is calculated according to the following formula:

[0077]

[0078] where A is the communication quality index between the vehicle and the RSU, N k is the total duration of disconnections and reconnections during the data transmission process, and RSSI is the signal strength. The square root form is used in the formula, indicating that the influence of the signal strength and disconnection duration on the communication quality is non-linear.

[0079] When N k increases, it means that the instability during the communication process is higher, resulting in a decrease in communication quality. Therefore, A will decrease accordingly; the higher the RSSI value, the stronger the received signal and the better the communication quality. Therefore, when RSSI increases, A will also increase; that is, it shows that RSSI and A are positively correlated, and N k and A are negatively correlated.

[0080] The specific logic for obtaining the computing resource index of the RSU is as follows: by referring to the hardware specification book and technical documents provided by the manufacturer, obtain the memory size and continuous power supply capacity of the RSU, and denote the memory size as G and the continuous power supply capacity as W,

[0081] After dimensionless processing the memory size and continuous power supply capacity, the computing resource index of the RSU is calculated according to the following formula:

[0082] B = (α * G + β * W) 2

[0083] Wherein, B is the computing resource index of the RSU, G is the memory size, W is the continuous power supply capacity, α and β are the proportionality coefficients of the memory size and the continuous power supply capacity respectively, β > α > 0, and satisfy: α + β = 1. This is because in the vehicle networking environment, the RSU usually needs to continuously provide services under various conditions, and the continuous power supply capacity is the key to ensuring the stable operation of the RSU. Especially under high load conditions, the RSU needs to have sufficient power to support its computing and communication functions, so a higher weight is given to it.

[0084] When G increases, it means that the RSU can store more data and execute more computing tasks, which can improve the efficiency and speed of data processing, and B increases accordingly; when W increases, it means that the RSU can maintain stable operation for a longer time. This is crucial for reliability under high load conditions, so B will increase accordingly; that is to say, G, W and B are positively correlated.

[0085] According to the moving trajectory roadmap of the vehicle, select the RSU closest to the vehicle currently as the best offloading node, and collect the distance between the user vehicle and this RSU, denoted as X;

[0086] After dimensionless processing of the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resources of the RSU, generate the revenue evaluation index of this RSU;

[0087] The formula for calculating the revenue evaluation index is as follows:

[0088]

[0089] Wherein, QS is the revenue evaluation index, used to evaluate the interaction revenue between the user vehicle and the RSU; A is the communication quality index between the vehicle and the RSU, B is the computing resource index of the RSU, and X is the distance between the user vehicle and this RSU. The logarithmic function is used in the formula, and the influence of the logarithmic function on extreme values is relatively small, which makes the model more robust in the face of noise and abnormal data.

[0090] When A increases, it means higher quality signals, lower latency and fewer packet losses, which will directly improve the user's communication experience, and QS will increase accordingly; higher computing resources can enhance the redundancy ability of the RSU, improve the reliability of the system, and reduce system failures or delays caused by insufficient resources. Therefore, when B increases, QS will also increase; as the distance increases, the signal strength may weaken, resulting in a decline in communication quality. Therefore, when X increases, QS will decrease; that is to say, A, B and QS are positively correlated, and X and QS are negatively correlated.

[0091] In step 2, by selecting the RSU closest to the vehicle as the optimal offloading node according to the vehicle's movement trajectory roadmap, a comprehensive evaluation mechanism is formed by combining multiple factors such as distance, communication quality, and computing resources. The advantage of this method is that it can dynamically select the most suitable RSU for task offloading, ensuring optimized resource utilization in complex traffic environments. Compared with existing technologies, traditional offloading strategies usually only rely on static distance judgment and fail to effectively consider the impact of communication quality and computing resources, resulting in low offloading success rate and increased latency.

[0092] In this solution, adopting step 2 can significantly improve the efficiency and success rate of task offloading. By comprehensively considering the distance, communication quality index, and computing resource index of the RSU, the system can flexibly respond to the dynamic changes of vehicles and the changing network environment, thus better supporting the task requirements of vehicles during driving. This optimization not only improves the data interaction efficiency between the vehicle and the RSU but also provides a more accurate basis for subsequent benefit evaluation and decision-making, overall improving the intelligent level and resource utilization efficiency of the vehicle networking system.

[0093] Step 3: Determine whether it meets the offloading requirements of the RSU according to the benefit evaluation index. When the offloading requirements of the RSU are not met, the RSU performs a reject offloading operation on the user vehicle and uses the queuing algorithm to sequentially select the remaining RSUs to determine whether they meet the offloading requirements until offloading is possible, achieving the Nash equilibrium between the user vehicle and the RSU;

[0094] In this embodiment, the benefit evaluation index is compared with a preset evaluation threshold:

[0095] If QS n ≥QY, it is determined that the current benefit evaluation index meets the offloading requirements of the RSU n The user vehicle selects the RSU n for task offloading. After the task is processed, the RSU n returns the calculation result to the user vehicle and calculates the total energy consumed by this task offloading;

[0096] If QS n <QY, it is determined that the current benefit evaluation index does not meet the offloading requirements of the RSU n The RSU performs a reject offloading operation on the user vehicle and uses the queuing algorithm to sequentially select the remaining RSUs to determine whether they meet the offloading requirements until offloading is possible;

[0097] Where QS n is the benefit evaluation index of the RSU n n is the index of the RSU, and QY is the preset evaluation threshold.

[0098] The queuing algorithm is specifically as follows: Define the user vehicle as V m , where m is the index of the user vehicle, define the amount of task data generated in each time slot as U m (t), and the task is temporarily stored locally in the user vehicle and unloaded to the RSU for computing and processing after the task is issued. The following data queue formula is formed in the user vehicle:

[0099] Q m (t + 1) = max{Q m (t) - L m (t) + U m (t), 0}

[0100] Among them, Q m (t + 1) represents the amount of data existing in the user vehicle V m in the (t + 1)-th time slot, L m (t) represents the throughput of the user vehicle V m in the t-th time slot, that is, the amount of data unloaded to the RSU, U m (t) represents the amount of newly generated data of the user vehicle V m in the t-th time slot;

[0101] Define the communication sub-channel bandwidth between the user vehicle V m and the roadside unit RSU n as B m,n , when the user vehicle V m selects the roadside unit RSU n for task offloading in the t-th time slot, calculate the amount of tasks unloaded by the user vehicle V m to the RSU n according to the following formula:

[0102]

[0103] Among them, z m,n,t represents the amount of tasks unloaded by the user vehicle V m to the RSU n , τ is a fixed transmission time, P is the transmission power, G m,n,t is the channel gain from the user vehicle V m to the RSU n in the t-th time slot, σ 2 is the noise power;

[0104] The calculation formula for the throughput of the user vehicle V m is as follows:

[0105]

[0106] Among them, Lm (t) represents the user vehicle V in the tth time slot m The throughput, X m,n,t Represents the user's vehicle V m In the tth time slot, the RSU n When X m,n,t =1, indicating that the user's vehicle V m Select RSU in the tth time slot n To offload tasks, if RSU n Reject task unloading, mark the task, and prohibit such tasks from being delivered to the roadside unit RSU n Unload and use the queuing algorithm to transfer the task to the next RSU for unloading until the unloading conditions are met.

[0107] Step 3 determines the RSU's unloading requirements based on the benefit evaluation index and uses a queuing algorithm to effectively manage the task unloading of user vehicles, significantly improving the system's flexibility and the success rate of task processing. The advantage of this method is that it can not only evaluate the availability and processing capacity of each RSU in real time, but also quickly adjust the unloading strategy in the event of unloading rejection and find other available RSUs for task processing. Compared with existing technologies, traditional methods often lack dynamic adjustment and redundant processing capabilities, resulting in unloading requests often failing due to rejection by a single RSU, affecting the efficiency of the overall system.

[0108] In this solution, adopting step 3 can effectively improve the interaction efficiency between the vehicle and the RSU and ensure that the task offloading can proceed smoothly. By achieving the Nash equilibrium between the user vehicle and the RSU, the vehicle can adjust the offloading strategy according to real-time feedback to minimize the delay and failure rate of task processing. This mechanism not only improves the success rate of the task, but also optimizes the utilization of system resources, so that the entire Internet of Vehicles environment can still operate efficiently in the face of dynamic changes, thereby enhancing the overall intelligence level and reliability of the system.

[0109] Step 4: After the RSU processing is completed, the calculation result is returned to the user vehicle, and the total energy consumed by this task offloading is calculated;

[0110] In this embodiment, after the RSU processing is completed, the calculation result is returned to the user vehicle, and the total energy consumed by this task unloading is calculated:

[0111] The total energy consumed by task offloading is calculated based on the following formula:

[0112] The total energy consumed by the task being offloaded to the RSU for the first time for:

[0113]

[0114] The total energy consumed when selecting the y-th RSU for offloading according to the queuing algorithm if the task is rejected by the RSU during the first offloading attempt is as follows:

[0115]

[0116] where τ is the fixed transmission time, P is the transmission power, and t R is the time consumed by the RSU to process the task, and t y represents the time required for data transmission from the user vehicle to the RSU when selecting the y-th RSU for task offloading. y is the index of the RSU, r represents the index of the RSU finally selected after multiple attempts when the user vehicle's first attempt to offload the task to the RSU is rejected, and ω represents the average task arrival rate, that is, the number of tasks sent by the user vehicle to the RSU per unit time.

[0117] Step 4 reflects an efficient information transfer mechanism and energy management ability by returning the calculation result to the user vehicle in a timely manner after the RSU processes the task and calculating the total energy consumed for this task offloading. The advantage of this method is that it not only ensures that the user vehicle can quickly obtain the required calculation result, thereby improving the user experience and system response speed, but also provides clear data on energy consumption to help users understand the energy consumption of task offloading. Compared with the existing technology, traditional offloading schemes often lack real-time monitoring and feedback of energy consumption, resulting in insufficient transparency of resource use for users.

[0118] In this solution, adopting Step 4 can effectively promote the optimized management of the overall task offloading scheme. Through real-time feedback of the calculation result and energy consumption, the user vehicle can perform more reasonable task scheduling and energy management based on this information, further improving the overall efficiency of the system. This cyclic information feedback mechanism not only enhances the collaboration ability between various nodes in the vehicle-to-everything (V2X) environment but also provides valuable data support for future intelligent transportation management, helping to continuously optimize and upgrade the task offloading strategy and improve the intelligence level of the entire V2X network.

[0119] Please refer to Figure 2 , a task offloading system based on regional RSUs in a vehicle-to-everything (V2X) environment, including:

[0120] A vehicle trajectory prediction module, configured to collect historical trajectory maps of vehicles, construct a vehicle trajectory prediction model based on a convolutional neural network and a path planning algorithm, calculate the optimal path according to the starting position and target position of the vehicle, and use the optimal path as the vehicle's moving trajectory roadmap;

[0121] The RSU selection module is used to select the RSU closest to the vehicle as the best offloading node according to the moving trajectory roadmap of the vehicle, obtain the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, and generate the revenue evaluation index of the RSU after dimensionless processing of the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU.

[0122] The offloading requirement judgment module is used to judge whether the offloading requirements of the RSU are met according to the revenue evaluation index. When the offloading requirements of the RSU are not met, the RSU performs a rejection offloading operation on the user vehicle, and uses the queuing algorithm to sequentially select the remaining RSUs to judge whether the offloading requirements are met until offloading is possible, so as to achieve the Nash equilibrium between the user vehicle and the RSU.

[0123] The result return and energy consumption calculation module is used to return the calculation result to the user vehicle after the RSU finishes processing, and calculate the total energy consumed by this task offloading.

[0124] The above formulas are all dimensionless and take their numerical calculations. The formula is a formula obtained by collecting a large amount of data for software simulation to get the closest real situation. The preset parameters in the formula are set by those skilled in the art according to the actual situation.

[0125] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0126] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units. They can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0127] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present application, and all should be covered by the protection scope of the present application.

Claims

1. A task offloading method based on regional RSU in a vehicle networking environment, characterized in that: The specific steps include: Step 1: Collect the historical trajectory map of the vehicle, build a vehicle trajectory prediction model based on the convolutional neural network and path planning algorithm, calculate the optimal path according to the starting position and target position of the vehicle, and use the optimal path as the vehicle's moving trajectory route map; Step 2: According to the vehicle's moving trajectory map, select the RSU that is currently closest to the vehicle as the best unloading node, obtain the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, and perform dimensionless processing on the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU to generate the benefit evaluation index of the RSU; Step 3: Determine whether the unloading requirements of the RSU are met according to the benefit evaluation index. If the unloading requirements of the RSU are not met, the RSU refuses to unload the user vehicle and uses the queuing algorithm to select the remaining RSUs in turn to determine whether they meet the unloading requirements until unloading is possible, thus achieving the Nash equilibrium between the user vehicle and the RSU. Step 4: After the RSU processing is completed, the calculation results are returned to the user vehicle, and the total energy consumed by this task offloading is calculated.

2. According to claim 1, a task offloading method based on regional RSU in a connected vehicle environment is characterized by: The vehicle trajectory prediction model is constructed based on a convolutional neural network and a path planning algorithm. The construction process of the vehicle trajectory prediction model is as follows: The historical trajectory map of the vehicle is collected and used as input data. The trained convolutional neural network is used to extract features from the preprocessed historical trajectory map. The topological structure of the road network is constructed based on the extracted features. The Dijkstra path planning algorithm is selected to calculate the optimal path based on the starting position and target position of the vehicle, and the optimal path is used as the vehicle's moving trajectory route map.

3. According to claim 1, a task offloading method based on regional RSU in a connected vehicle environment is characterized by: The communication quality index between the vehicle and the RSU is obtained based on the following specific logic: within T time, the vehicle's wireless communication module is used to collect the signal strength of multiple communications between the vehicle and the RSU, and the average is calculated as the signal strength when the vehicle and the RSU communicate, which is recorded as RSSI; The user vehicle sends a test data packet to the RSU, obtains the number of disconnections and reconnections during data transmission within T time and the corresponding duration of disconnection and reconnection, and calibrates the duration corresponding to the pth disconnection and reconnection within T time as Tp. Then, the total duration of disconnection and reconnection during data transmission within T time is N k The expression to be calculated is: Wherein, p represents the number of disconnections and reconnections during data transmission within T time, p=1, 2, 3, 4, ..., q, q is the total number of disconnections and reconnections within T time; To N k The RSSI is dimensionless, and the communication quality index between the vehicle and the RSU is calculated according to the following formula: Among them, A is the communication quality index between the vehicle and RSU, N k It is the total time of disconnection and reconnection during data transmission, and RSSI is the signal strength.

4. According to claim 1, a task offloading method based on regional RSU in a connected vehicle environment is characterized by: The specific logic for obtaining the computing resource index of the RSU is as follows: by consulting the hardware specifications and technical documents provided by the manufacturer, the memory size and continuous power supply capacity of the RSU are obtained, and the memory size is recorded as G and the continuous power supply capacity is recorded as W. After non-dimensionalizing the memory size and continuous power supply capability, the computing resource index of the RSU is calculated according to the following formula: B=(α*G+β*W) 2 Among them, B is the computing resource index of RSU, G is the memory size, W is the continuous power supply capability, α and β are the proportional coefficients of memory size and continuous power supply capability respectively, β>α>0, and satisfies: α+β=1.

5. According to claim 1, a task offloading method based on regional RSU in a connected vehicle environment is characterized by: According to the vehicle's moving trajectory map, the RSU closest to the vehicle is selected as the best unloading node, and the distance between the user's vehicle and the RSU is collected, recorded as X; After dimensionless processing of the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resources of the RSU, the benefit evaluation index of the RSU is generated; The formula for calculating the return assessment index is as follows: Among them, QS is the benefit evaluation index, A is the communication quality index between the vehicle and the RSU, B is the computing resource index of the RSU, and X is the distance between the user vehicle and the RSU.

6. According to claim 1, a method for task offloading based on regional RSU in a connected vehicle environment is characterized by: Compare the benefit assessment index with the preset assessment threshold: If QS n ≥QY, judging that the current income evaluation index meets the RSU n The user vehicle selects RSU n After the task is completed, RSU n The calculation result is returned to the user's vehicle, and the total energy consumed by this task unloading is calculated; If QS n <QY, it is determined that the current revenue evaluation index does not meet the offloading requirements of the RSU n The RSU performs a reject offloading operation on the user's vehicle, and uses the queuing algorithm to sequentially select the remaining RSUs to determine whether they meet the offloading requirements until offloading is possible; In the formula, QS n For RSU n The benefit evaluation index is , n is the index of RSU, and the index is assigned to each RSU in order from the nearest to the farthest distance from the vehicle, and QY is the preset evaluation threshold.

7. The method for offloading tasks based on regional RSU in a connected vehicle environment according to claim 6 is characterized by: The queuing algorithm is specifically as follows: define the user vehicle as V m , where m is the index of the user vehicle, and the amount of task data generated in each time slot is defined as U m (t), and the task is temporarily stored in the user's vehicle. After the task is issued, it is unloaded to the RSU for calculation and processing. The user's vehicle forms the following data queue formula locally: Q m (t+1)=max{Q m (t)-L m (t)+U m (t),0} Among them, Q m (t+1) indicates that there is a user vehicle V in the t+1th time slot m The amount of data on L m (t) represents the user vehicle V in the tth time slot m The throughput, i.e. the amount of data offloaded to the RSU, U m (t) represents the user vehicle V in the tth time slot m The amount of new data generated; Define user vehicle V m and Roadside Units RSU n The bandwidth of the communication subchannel between m,n , when the user's vehicle V m Select the roadside unit RSU in the tth time slot n When unloading tasks, the user vehicle V is calculated according to the following formula: m Offloading to RSU n Amount of tasks: Among them, z m,n,t Represents the user's vehicle V m Offloading to RSU n The task volume, τ is the fixed transmission time, P is the transmission power, G m,n,t is the user vehicle V in the tth time slot m To RSU n The channel gain, σ 2 is the noise power; User Vehicle V m The throughput is calculated as follows: Among them, L m (t) represents the user vehicle V in the tth time slot m The throughput, X m,n,t Represents the user's vehicle V m In the tth time slot, the RSU n When X m,n,t =1, indicating that the user's vehicle V m Select RSU in the tth time slot n To offload tasks, if RSU n Reject task unloading, mark the task, and prohibit such tasks from being delivered to the roadside unit RSU n Unload and transfer the task to the next RSU for unloading using the queuing algorithm in the order of index until the unloading condition is met. N is the number of all RSUs.

8. The method for offloading tasks based on regional RSU in a connected vehicle environment according to claim 7 is characterized by: After the RSU processing is completed, the calculation results are returned to the user vehicle, and the total energy consumed by this task offloading is calculated: The total energy consumed by task offloading is calculated based on the following formula: The total energy consumed by the task being offloaded to the RSU for the first time for: If the task is rejected by the RSU when it is unloaded for the first time, the total energy consumed when the yth RSU is selected for unloading according to the queuing algorithm for: Among them, τ is the fixed transmission time, P is the transmission power, t R The time consumed by RSU to process the task, t n represents the time required for data transmission from the user vehicle to the RSU when the nth RSU is selected for task offloading, r represents the index of the RSU finally selected by the user vehicle, and y∈[1,N], ω represents the average task arrival rate, that is, the number of tasks sent to the RSU by the user vehicle per unit time.

9. A task offloading system based on regional RSU in a connected vehicle environment, characterized by: The task offloading system based on regional RSU in a connected vehicle environment is used to execute the task offloading method based on regional RSU in a connected vehicle environment according to any one of claims 1 to 8, including: The vehicle trajectory prediction module is used to collect the historical trajectory map of the vehicle, build a vehicle trajectory prediction model based on the convolutional neural network and path planning algorithm, calculate the optimal path according to the starting position and target position of the vehicle, and use the optimal path as the vehicle's moving trajectory route map; The RSU selection module is used to select the RSU closest to the vehicle as the best unloading node according to the vehicle's moving trajectory route map, obtain the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU, and generate the benefit evaluation index of the RSU after dimensionless processing of the distance between the vehicle and the RSU, the communication quality index between the vehicle and the RSU, and the computing resource index of the RSU. The unloading requirement judgment module is used to judge whether the unloading requirement of the RSU is met according to the benefit evaluation index. When the unloading requirement of the RSU is not met, the RSU refuses to unload the user vehicle and uses the queuing algorithm to select the remaining RSUs in turn to judge whether they meet the unloading requirements until unloading is possible, thereby achieving the Nash equilibrium between the user vehicle and the RSU; The result return and energy consumption calculation module is used to return the calculation result to the user vehicle after the RSU processing is completed, and calculate the total energy consumed by this task unloading.

Images