Incentive System and Method for Cooperative Computing in Intelligent Connected Vehicles

Abstract

This invention relates to the technical field of edge computing combined with artificial intelligence, proposing an incentive system and method for collaborative computing in intelligent connected vehicles. It integrates the computing resources of idle vehicles into the task offloading network of the vehicle network, utilizing idle vehicles for collaborative computing, further improving vehicle resource utilization and effectively reducing the total system cost. The invention introduces an incentive system that quantifies offloading benefits through points, guiding the offloading decisions of computing tasks among idle vehicles, fully mobilizing the enthusiasm of idle vehicles to participate in task offloading, optimizing the allocation of computing resources among vehicles, and improving the efficiency of task offloading in the vehicle network. Furthermore, the invention integrates multi-agent collaborative learning into a cloud-edge-device collaborative architecture, placing the experience pool on a cloud server, enabling comprehensive acquisition of global training samples, enhancing the global relevance of decisions, and further improving model performance.

Description

Technical Field

[0001] This invention relates to the technical field of edge computing and artificial intelligence, specifically to an incentive system and method for collaborative computing in intelligent connected vehicles. Background Technology

[0002] The application and promotion of cloud-edge-device collaborative architecture in intelligent connected vehicles has enhanced vehicle intelligence and service capabilities, while also placing higher demands on real-time data processing. Traditional cloud-edge-device collaborative computing relies on edge servers as the main computing entity for task offloading, failing to fully consider that as vehicles are updated and iterated, they themselves possess certain surplus computing resources capable of participating in task offloading. However, vehicles are client-side components, and participating in task offloading requires additional computing and communication resources, resulting in low enthusiasm for vehicles to participate in collaborative computing and limiting the utilization of surplus resources within the vehicle fleet.

[0003] Incentives, originating from management science, refer to the methods organizations use to stimulate and guide the behavior of their members, thereby achieving organizational and individual goals. Introducing a fair and reasonable incentive system into the existing vehicle task unloading system, quantifying the vehicle's contribution to time delay and energy consumption during task unloading using methods such as points, and formulating relevant rules to incentivize participating vehicles, further enhances the vehicles' proactive performance during task unloading and ensures optimal resource utilization among vehicles. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide an incentive system and method for collaborative computing of intelligent connected vehicles in the context of idle vehicles participating in task offloading, so as to improve the enthusiasm of vehicles to participate in task offloading.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] An incentive system for collaborative computing in intelligent connected vehicles, comprising a cloud server layer, an edge server layer, and a vehicle layer;

[0007] The vehicle layer includes an information collection module, an information upload module, an unloading request module, an unloading execution module, and a data storage module;

[0008] The system comprises the following modules: an information acquisition module for obtaining vehicle information, including real-time location, task queue latency, and current score, which are essential for decision-making; an information upload module for uploading vehicle information to the information receiving module at the edge server layer; an unloading request module for sending an unloading request to the incentive decision module at the edge server layer; an unloading execution module for executing the unloading decision from the incentive decision module at the edge server layer; vehicles receiving unloading decisions and responding vehicles directly transmit and process tasks, and update vehicle score information based on the unloading results; and a data storage module for storing vehicle task calculation data and results, score information, and other necessary vehicle information.

[0009] The edge server layer includes an information receiving module, an information maintenance module, an incentive decision-making module, and a sample uploading module;

[0010] The system comprises the following modules: an information receiving module for receiving vehicle information within the communication range of the edge server; an information maintenance module for maintaining an idle vehicle information table, which updates the table in real time based on the idle vehicle information received by the information receiving module; an incentive decision-making module for incentivizing the learning and reasoning of the decision-making model; upon receiving an unloading request, the module combines the vehicle information table from the information maintenance module to make a decision, selecting the best responding vehicle based on maximizing the integral; after the decision is made, the module transmits the acquired local samples to the sample upload module; upon receiving global samples from the cloud server, the module trains and updates the model based on the global samples; and a sample upload module for uploading local samples generated during the decision-making process to the sample receiving module at the cloud server layer.

[0011] The cloud server layer is divided into a sample receiving module, a sample distribution module, a sample integration module, and a sample storage module;

[0012] The sample receiving module is used to receive local samples uploaded by the edge server; the sample distribution module is used to periodically distribute global samples to the incentive decision module of the edge server; the sample integration module is used to integrate local samples into global samples; the sample storage module is used to store local samples and global samples, and assumes the function of the experience pool. It receives local samples from the sample receiving module, sends the local samples to the sample integration module for processing into global samples, and periodically sends the global samples to the sample distribution module.

[0013] An incentive method for cooperative computing in intelligent connected vehicles includes the following steps:

[0014] Step S1: Construct an incentive system for collaborative computing of intelligent connected vehicles. The system includes a three-layer architecture: cloud server layer, edge server layer, and vehicle layer.

[0015] In step S2, the vehicle layer synchronizes vehicle information and points information with the edge server layer, and the edge server layer maintains the vehicle information table based on the received vehicle information.

[0016] Step S3: The vehicle layer generates a task. When the vehicle itself cannot complete the task within the time limit, it sends an unload request to the edge server layer. The edge server layer makes an unload decision through the incentive decision model and sends it back to the vehicle layer. The vehicle layer completes the task unloading based on the unloading decision and updates the scores of the vehicles that participated in the unloading.

[0017] In step S4, after making a decision, the edge server layer generates local samples and uploads them to the experience pool of the cloud server layer. The cloud server integrates the local samples in the experience pool into global samples and then distributes them to the incentive decision model of the edge server layer for model training.

[0018] The beneficial results of this invention are as follows:

[0019] (1) This invention integrates the computing resources of idle vehicles into the task offloading network of the Internet of Vehicles, and uses idle vehicles for collaborative computing, thereby further improving the resource utilization rate of vehicles and effectively reducing the total cost of the system.

[0020] (2) The present invention introduces an incentive system with points as a medium to quantify the unloading benefits, which can guide the unloading decision of computing tasks among idle vehicles, fully mobilize the enthusiasm of idle vehicles to participate in task unloading, optimize the allocation of computing resources among vehicles, and improve the unloading efficiency of vehicle networking tasks.

[0021] (3) This invention integrates multi-model collaborative learning into a cloud-edge-device collaborative architecture, places the experience pool on the cloud server, can comprehensively obtain global training samples, enhance the global correlation of decision-making, and further improve model performance. Attached Figure Description

[0022] Figure 1 This is a system architecture diagram for collaborative computing of intelligent connected vehicles in an embodiment of the present invention.

[0023] Figure 2 This is an architecture diagram of the vehicle layer in an embodiment of the present invention.

[0024] Figure 3 This is an architecture diagram of the edge server layer in an embodiment of the present invention.

[0025] Figure 4 This is an architecture diagram of the cloud server layer in an embodiment of the present invention. Detailed Implementation

[0026] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings.

[0027] like Figure 1As shown, this invention proposes an incentive system for collaborative computing in intelligent connected vehicles. The system architecture is divided into a cloud server layer, an edge server layer, and a vehicle layer.

[0028] The vehicle layer is divided into an information acquisition module, an information upload module, an unloading request module, an unloading execution module, and a data storage module. The information acquisition module obtains vehicle information, including real-time location, task queue latency, and current score, necessary for decision-making. The information upload module uploads vehicle information to the information receiving module of the edge server layer, determining whether to upload based on whether the vehicle's computing resources are idle; information is only uploaded when the vehicle has no computing tasks and is idle. The unloading request module sends unloading requests to the incentive decision module of the edge server layer, based on the vehicle's current task queue and the computational load of the new task. If the vehicle's own computing power is insufficient to meet the latency requirements of the new task, it initiates an unloading request to the edge server layer. The unloading execution module executes the unloading decisions from the incentive decision module in the edge server layer. Upon receiving an unloading decision request, the requesting vehicle and the responding vehicle directly transmit and process the task, and update the vehicle's score information based on the unloading result. The data storage module stores vehicle task calculation data and results, score information, and other necessary vehicle information.

[0029] The edge server layer is divided into an information receiving module, an information maintenance module, an incentive decision-making module, and a sample uploading module. The information receiving module receives vehicle information within the communication range of the edge server; the information maintenance module maintains an idle vehicle information table, updating it in real time based on the idle vehicle information received by the information receiving module; the incentive decision-making module is used to incentivize the learning and inference of the incentive decision-making model. Upon receiving an unloading request, it combines the vehicle information table from the information maintenance module to make a decision, selecting the best responding vehicle based on maximizing the integral. After the decision is completed, it transmits the acquired local samples to the sample uploading module. Upon receiving global samples from the cloud server, it trains and updates the model based on the global samples; the sample uploading module uploads local samples generated during the decision-making process to the sample receiving module of the cloud server layer.

[0030] The cloud server layer is divided into a sample receiving module, a sample distribution module, a sample integration module, and a sample storage module. The sample receiving module receives local samples uploaded from edge servers; the sample distribution module periodically distributes global samples to the incentive decision-making module of the edge servers; the sample integration module integrates local samples into global samples. It splices and integrates local samples from different edge servers in the same time slot, unifying scattered local information into global samples that represent global information; the sample storage module stores local and global samples, acting as an experience pool. It receives local samples from the sample receiving module, sends them to the sample integration module for processing into global samples, and periodically sends the global samples to the sample distribution module.

[0031] Vehicle information is obtained from other modules and the vehicle's own sensors.

[0032] Vehicle credits are stored in the vehicle's data storage module using Elliptic Curve Cryptography (ECC), which offers higher security performance with the same number of bits compared to the commonly used RSA encryption algorithm.

[0033] Idle vehicles are defined as vehicles with idle computing resources, that is, vehicles whose task queues are empty and which have not generated new tasks.

[0034] The specific incentive method based on the above system is as follows:

[0035] Step S1: Construct an incentive system for collaborative computing in intelligent connected vehicles. The system consists of a three-layer architecture: a cloud server layer, an edge server layer, and a vehicle layer.

[0036] In step S2, the vehicle layer synchronizes vehicle information and points information with the edge server layer, and the edge server layer maintains the vehicle information table based on the received vehicle information.

[0037] In step S3, the vehicle layer generates a task. If a vehicle cannot complete the task within the time limit, it sends an unloading request to the edge server layer. The edge server layer makes an unloading decision using an incentive decision model and sends it back to the vehicle layer. The vehicle layer completes the task unloading based on the unloading decision and updates the scores of the vehicles involved in the unloading.

[0038] In step S4, after making a decision, the edge server layer generates local samples and uploads them to the experience pool of the cloud server layer. The cloud server integrates the local samples in the experience pool into global samples and then distributes them to the incentive decision model of the edge server layer for model training.

[0039] In step S1, the cloud server layer contains one cloud server, the edge server layer contains K edge servers, and the vehicle layer contains N vehicles. The sets of K edge servers and N vehicles are represented as follows:

[0040] K={1,2,...,K},N={1,2,...N}

[0041] The cloud server exchanges data with the edge server via a dedicated line, connecting all the edge servers; the edge server forwards data through the roadside unit (RSU) and indirectly exchanges data with vehicles within a certain range, undertaking the tasks of information integration and offloading decisions; vehicles exchange data directly with each other via wireless communication.

[0042] The agreed-upon task offloading occurs on the edge server e l Within the communication range, during the task unloading process, the vehicle that initiates the task unloading is called the requesting vehicle, denoted as vehicle v. i The vehicle that responds to the unloading request is called the responding vehicle, denoted as vehicle v. k .

[0043] The communication strength between vehicles depends on the transmit power, channel bandwidth, noise level, and the communication interference from other vehicles. According to Shannon's formula, vehicle v... i With vehicle v k The transmission rate during wireless communication can be expressed as:

[0044]

[0045] Among them B i,k with h i.k These represent vehicles v. i With vehicle v k The channel bandwidth and channel gain between P i,k Indicates vehicle v i To vehicle v k The transmit power during data transmission, where N0 represents the channel noise power, and P... i,a with h i,a These represent vehicles v. i With third-party vehicles v a The transmit power and channel gain between them.

[0046] In step S2, the vehicle information table is used to maintain information on idle vehicles within the communication range of the edge server and to provide necessary decision-making information for the incentive decision-making model. The specific definition of idle vehicles in this embodiment is as follows:

[0047] Idle vehicles refer to vehicles that are not currently performing computational tasks. Specifically, they can be defined as vehicles whose current task processing latency is less than τ. w The vehicles, where the time interval τ w τ represents the queuing delay threshold. To ensure that vehicles can process unloading tasks without queuing delay, τ... w The value of is not greater than the processing delay t of the incentive decision-making process.p .

[0048] To ensure the dynamic maintenance of the vehicle information table, the vehicle information upload module needs to impose constraints on both communication strength and the availability of idle vehicles before uploading information. Only idle vehicles within the effective transmission range are allowed to upload information.

[0049] The constraint on communication strength can be simplified to the Euclidean distance limit between the vehicle and the RSU. Define the vehicle v. i The communication distance threshold with RSU is Its value is related to vehicle v i Maximum transmit power P i Tmax Positive correlation. Edge server e l The corresponding RSU position is set to Vehicle real-time location set The specific representation of the constraint is as follows:

[0050]

[0051] The constraint on idle vehicles, based on the above definition, can be transformed into a limitation on the processing latency of the vehicle's current task. Let the processing latency of the vehicle's current task be set as τ. p The specific representation of the constraint is as follows:

[0052] τ p ≤τ w

[0053] like Figure 2 As shown, the vehicle information table is deployed on the information maintenance module of the edge server layer, and the edge server e l Receiving vehicle v i The specific process for obtaining and maintaining the vehicle information table is as follows:

[0054] Step S2-1, vehicle v i The information acquisition module obtains the vehicle's current credits via a bus and a data storage module. i This, along with other necessary information, is then transmitted to the information upload module.

[0055] Step S2-2, idle vehicles v that satisfy the information upload constraints i Upload your own information in real time via the information upload module. To the corresponding edge server e l The information receiving module in the middle, where ID i For vehicle v i Identifier, P i Tmax f is the vehicle's maximum transmission power. i Mv is the maximum processing frequency of the vehicle's Central Processing Unit (CPU). i For vehicle speed, This indicates the vehicle's real-time location.

[0056] Step S2-3, Edge Server e l The information maintenance module obtains information from the information receiving module. i Then, in information I i Add current timestamp at the end Receive information

[0057]

[0058] Step S2-4: The information maintenance module removes outdated information from the vehicle information table. An agreed time difference tolerance threshold τ is set. tag If the difference between the current timestamp and the information timestamp is the maximum allowed value, then information I is discarded. j The basis for ' is expressed as follows:

[0059]

[0060] Step S2-5, the information maintenance module, based on information I i Vehicle Identifier ID i Check if the vehicle information exists. If the vehicle information exists, update the existing information; otherwise, add the new vehicle information to the table.

[0061] like Figure 3 As shown, the incentive decision model is deployed on the incentive decision module of the edge server layer. The specific steps for the vehicle layer and the edge service layer to use the incentive decision model to offload tasks are as follows:

[0062] Step S3-1, vehicle v i The unload request module on the top sends a request to the edge server e l Initiate uninstallation request R j (ID i ,d j ,c j ,T j ), where d j For the amount of task data, c j T is the number of CPU cycles required for the task. j This represents the maximum allowable delay for the task.

[0063] Step S3-2, Edge server e l The incentive decision module received the unload request R j Then, by using the incentive decision-making model and the idle vehicle information table of the information maintenance module, the responding vehicle v is determined with the objective of maximizing the integral. kAt the same time, determine the requesting vehicle v i integral increment With response vehicle v k integral increment

[0064] Step S3-3, the corresponding edge server e l The incentive decision-making module will determine the decision results. Return request vehicle v i With response vehicle v k .

[0065] Steps S3-4, vehicle v i The unloading execution module receives the decision result Re j Then, towards vehicle v k The unloading execution module transmits the computational tasks stored in the data storage module and updates the vehicle credits.

[0066] Steps S3-5, vehicle v k The unloading execution module receives the decision result Re j Afterwards, receive vehicle v i The system performs the calculations and processes the data, updating the vehicle's credits. After completion, the results are returned to the vehicle v. i The uninstallation execution module completes the uninstallation task.

[0067] To address the aforementioned task unloading process, this embodiment proposes a latency and energy consumption model for task unloading between vehicles. The specific analysis of the model is as follows:

[0068] Task R j Vehicle v i Unload to vehicle v k The process involves five steps: unloading request, incentive decision, data transmission, data processing, and data feedback. The latency and energy consumption of the unloading request and data feedback processes are negligible due to the small amount of data transmitted; the incentive decision latency is set to t. p Energy consumption is irrelevant to the vehicle and can be ignored; during data transmission, the latency is... Energy consumption is During data processing, the latency is Energy consumption is Where, d j With c j Task R j The amount of data and the number of CPU cycles required, f j,k For vehicle v k Computation task R j CPU processing frequency, P i,k Indicates vehicle v i To vehicle vk Transmit power during data transmission, r i,k For vehicle v i With vehicle v k The data transfer rate between them, where k is the effective switched capacitor, is related to the chip structure.

[0069] For the incentive decision-making model, this embodiment also proposes an optimization model with the objective of maximizing the integral, when vehicle v i To the edge server e l The incentive decision module issues the unloading task R j After the uninstallation request is received, the specific modeling process is as follows:

[0070] Step a, preset the processing delay t of the incentive decision-making process. p Its value is fixed, and it is taken as the average time delay of decision-making during the training process.

[0071] Step b, preset the integration parameter ρ t The unit benefit representing time savings is then used to incentivize vehicle v during the decision-making process. i The obtained delay benefit integral is expressed as This refers to the remaining task timeframe after the task has been unloaded, including the stimulus decision delay, data transmission delay, and data computation delay. Where T... j For task R j Maximum allowable delay, t p To incentivize decision-making processes to handle delays.

[0072] Step c, preset the integration parameter ρ e Representing the unit cost of energy consumption, the vehicle v in the incentive decision-making process i The required launch energy cost integral can be expressed as: Vehicle v k The energy consumption cost required to process the data is integraled as follows:

[0073] Step d: Preset the response incentives for incentive decision-making. This represents the points incentive that the responding vehicle can obtain after accepting the unloading request, and it is agreed that this incentive is related to the unloading task R. j The amount of data is positively correlated, that is... Where ρ r The incentive parameter is used to adjust the degree of incentive for task unloading.

[0074] Step e: Based on the above analysis, request vehicle v i The integral change generated during the incentive decision-making process is Response vehicle v k The integral change generated during the incentive decision-making process is

[0075] Step f: It is agreed that there are N vehicles in the system, and the number of tasks generated during the process is M, with v vehicles. i Uninstall task R j The points obtained are s i,j Vehicle v i The maximum transmission power is P i Tmax Vehicle v k The maximum CPU processing frequency is Task R j The maximum allowable delay is T j The optimization problem Q is established with the goal of maximizing the integral:

[0076]

[0077] Among them, constraint C1 restricts the scope of vehicle computing resources used, constraint C2 limits the range of vehicle transmission power, and constraint C3 restricts the task completion time limit.

[0078] like Figure 4 As shown, the experience pool is deployed on the sample storage module of the cloud server layer, and the specific process of sample management for the incentive decision model is as follows:

[0079] Step S4-1, the corresponding edge server e l The incentive decision module will assign task R j The unloading decision results are integrated into local training samples and transmitted to the sample receiving module of the cloud server through the sample upload module.

[0080] In step S4-2, the cloud server's sample receiving module stores the received local samples into the sample storage module.

[0081] In step S4-3, the cloud server's sample integration module integrates the local samples in the sample storage module into global samples and sends them back to the sample storage module.

[0082] In step S4-4, the cloud server's sample distribution module periodically samples from the sample storage module and distributes the sampled global samples to the edge server's incentive decision module.

[0083] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. Any equivalent modifications or variations made by those skilled in the art based on the content disclosed in the present invention should be included within the scope of protection set forth in the claims.

Claims

1. An incentive system for intelligent vehicle connected cooperative computing, characterized in that: The incentive system comprises a cloud server layer, an edge server layer and a vehicle layer; The vehicle layer comprises an information acquisition module, an information uploading module, an unloading request module, an unloading execution module and a data storage module; The information acquisition module is configured to acquire vehicle self-information, which is necessary for decision-making, such as real-time position, task queue delay and current score of the vehicle; the information uploading module is configured to upload vehicle information to the information receiving module of the edge server layer; the unloading request module is configured to send an unloading request to the incentive decision module of the edge server layer; the unloading execution module is configured to execute the unloading decision from the incentive decision module of the edge server layer; the vehicle receiving the unloading decision and the response vehicle directly perform task transmission and processing, and update the score information of the vehicle according to the unloading result; and the data storage module is configured to store vehicle task calculation data and calculation results, score information and other necessary information of the vehicle. The edge server layer comprises an information receiving module, an information maintaining module, an incentive decision module and a sample uploading module; The information receiving module is configured to receive vehicle information within the communication range of the edge server; the information maintaining module is configured to maintain an idle vehicle information table, which is updated in real time according to the idle vehicle information received by the information receiving module; the incentive decision module is configured to learn and infer the incentive decision model; when receiving the unloading request, it makes a decision in combination with the vehicle information table of the information maintaining module, and selects the best response vehicle according to the maximum score; after the decision is made, it transmits the obtained local sample to the sample uploading module; when receiving the global sample from the cloud server layer, it updates the model according to the global sample; and the sample uploading module is configured to upload the local sample generated in the decision-making process to the sample receiving module of the cloud server layer. The cloud server layer comprises a sample receiving module, a sample issuing module, a sample integrating module and a sample storage module; The sample receiving module is configured to receive the local sample uploaded by the edge server layer; the sample issuing module is configured to issue the global sample to the incentive decision module of the edge server layer at regular intervals; the sample integrating module is configured to integrate the local sample into the global sample; and the sample storage module is configured to store the local sample and the global sample, and functions as an experience pool, which receives the local sample from the sample receiving module, processes the local sample into the global sample by the sample integrating module, and sends the global sample to the sample issuing module at regular intervals. 2.The intelligent network connected vehicle cooperative computing oriented incentive system according to claim 1, characterized in that: The information uploading module uploads vehicle information according to whether the vehicle computing resource is in an idle state; only when the vehicle has no computing task and is in an idle state, the vehicle information is uploaded; an idle vehicle is defined as a vehicle whose computing resource is idle, i.e., the vehicle task queue is empty and no new task is generated by the vehicle. 3.The intelligent vehicle cooperative computing oriented incentive system according to claim 1, characterized in that: The unloading request module sends an unloading request according to the current task queue and the calculation amount of the new task of the vehicle; if the computing capacity of the vehicle itself is insufficient to meet the delay requirement of the new task, the unloading request is initiated to the edge server layer. 4.The intelligent vehicle cooperative computing oriented incentive system according to claim 1, wherein: The vehicle score is saved in the data storage module of the vehicle in an encrypted manner using the elliptic curve encryption algorithm (ECC). 5.The intelligent vehicle cooperative computing oriented incentive system according to claim 1, wherein: In the sample integration module, the local samples of different edge servers in the same time slot are spliced and integrated, and the scattered local information is integrated into global samples representing global information.

6. An incentive method using the intelligent vehicle collaboration computing oriented incentive system according to any one of claims 1-5, characterized in that: The method specifically comprises the following steps: Step S1, an incentive system for intelligent networked vehicle collaborative computing is constructed, and the system comprises a cloud server layer, an edge server layer and a vehicle layer in a three-layer architecture; Step S2, the vehicle layer synchronizes vehicle information and integral information to the edge server layer, and the edge server layer maintains a vehicle information table through the received vehicle information; Step S3, the vehicle layer generates a task, and when the vehicle itself cannot complete the task within a time limit, sends an offloading request to the edge server layer; The edge server layer makes an offloading decision through an incentive decision model and returns it to the vehicle layer; the vehicle layer completes task offloading according to the offloading decision and updates the integral of the vehicle participating in offloading; Step S4, the edge server layer generates a local sample after decision-making and uploads it to the experience pool of the cloud server layer, and the cloud server integrates the local samples in the experience pool into global samples and then issues them to the incentive decision model of the edge server layer for model training.

7. The method of energizing of claim 6, wherein: The cloud server layer in step S1 comprises one cloud server, the edge server layer comprises K edge servers, and the vehicle layer comprises N vehicles; the sets of K edge servers and N vehicles are represented as: The cloud server exchanges data with the edge servers through a dedicated line, and is connected to all the edge servers; the edge servers exchange data with vehicles within a radius through a roadside unit RSU, and bear the tasks of information integration and unloading decision R k The vehicles exchange data indirectly through wireless communication within a radius, and bear the tasks of information integration and unloading decision.

8. The method of energizing of claim 6, wherein: The vehicle information table in step S2 is used to maintain the idle vehicle information within the communication range of the edge server and provide necessary decision information for the incentive decision model, and the vehicle information is dynamically maintained by adding a time stamp to the vehicle information.

9. The method of energizing of claim 6, wherein: The incentive decision model in step S3 is a task offloading model between vehicles, which maximizes the integral as the optimization objective and selects the most suitable offloading response vehicle; the integral quantifies the contribution and benefit of the vehicle in the task offloading process, and the requesting vehicle obtains the time delay benefit of task offloading but needs to bear the energy cost of data transmission in the offloading process; the response vehicle needs to bear the data calculation energy cost, and the response integral positively related to the task data volume is set to motivate the response vehicle to make up for the calculation energy cost.

10. The method of energizing of claim 6, wherein: The experience pool in step S4 is used to store the training samples of the incentive decision model, the local samples are uploaded by the edge server layer, and the global samples are integrated from the local samples of different edge servers in the same time slot.

Images