Research On Vehicle Computation Offloading And Computation Resource Scheduling In VANET

With the rapid development of economy and science and technology,people’s lives have been improving recently.In the field of public transportation the number of motor vehicles has been increasing dramatically.But this has also brought a series of problems,such as traffic jams and frequent traffic accidents.VANET and its smart applications are effective solutions for these problems.For example,autonomous driving can avoid traffic accidents caused by driver fatigue,and can also avoid urban traffic jams through reasonable path planning.These intelligent applications have many types,such as delay-sensitive and delayinsensitive according to the time,resource-intensive and non-resource-intensive according to the requirements of storage resources.And many tasks can be split into small pieces of task.Most these applications have a common feature,that is very sensitive to the delay.Various applications need to process a large amount of data in a short period of time.This requires strong power of computing,but vehicular resources are not sufficient for these high requirements.Therefore,better methods for computing offload and computing resource allocation are needed.This article first considers the scenario of insufficient MEC computing power,and the vehicle’s own computing resources are limited,so complex computation tasks require other equipment with enough computing resources to help the vehicle to complete the calculation.To solve this problem,this paper takes the surrounding vehicles and its own resources as a resource pool,and uses all available on-board resources to provide low-latency services for intelligent applications.Based on the available resources of the resource pool,taking full account of the high-speed mobility of the vehicle and the instability of the communication link,this paper designs a single-task distributed computation offloading method,by splitting the complex computing task into multiple small tasks and assigning them to the surrounding vehicles for calculation,to obtain a lower calculation task processing delay.In the design of the allocation algorithm,this paper designs the optimal allocation scheme of resource allocation based on genetic algorithm and DQN respectively,and obtains good experimental results,greatly reducing the processing delay of complex tasks.The simulation proves that the model proposed in this paper can make full use of the computing resources of the surrounding vehicles,considering the mobility of the vehicle,the delay of communication transmission,and the separability of the task,which greatly reduces the execution time of the computing task.Secondly,in the scene of the motorcade,the two vehicles on the front and rear take too many calculation tasks and consume more computing energy,while vehicles in the motorcade only need to follow the vehicle before them,and their energy consumptions are very little,which leads to a problem that resources are not used fairly.For this problem,an online computation offload method based on DDPG is proposed.This is a joint offloading strategy that integrates all available resources around the vehicle,locally available computing resources,surrounding idle computing resources,and MEC-provided computing resources,offloading through three different calculations: local computing,V2 V,and V2 I,to complete the calculation of tasks in a collaborative way.This article is based on the long-term online computation offloading method designed by the DDPG algorithm,which uniformly schedules three computing resources and learns to obtain higher system utilization.This method maximizes the utilization of system resources while meeting the requirements of application delay.While ensuring the delay of computing tasks,it improves the utilization rate of the system,distributes the computing tasks to each device in a balanced manner,and provides a new solution for computing offloading.Simulation proves the effectiveness of the scheme.The algorithm can converge and achieve the best results under different numbers and sizes of computing tasks.