Research On Task Offloading Strategy In Vehicular Cloud Computing

Vehicular cloud computing is a hot research area of the internet of vehicle technologies.Computation offloading is the key part of vehicular cloud computing.It is an effective technology to balance between tasks’ diverse requirements and the limited computing ability on vehicular computers.This thesis takes vehicular cloud as the research background and focuses on the task offloading strategy of vehicular cloud computing.On the basis of summarizing the previous task offloading technology,the task offloading strategy in vehicular cloud computing is deeply studied.The research contents are summarized as follows from the perspectives of resource allocation and task allocation respectively:1.Existing game-based resource allocation strategies in vehicular cloud computing considered that the price of the task is fixed.The impact of task urgency and network conditions on prices is ignored,so that the resources of tasks with a tight execution time or busy surrounding networks cannot be guaranteed during the game.To deal with this problem,an offloading node selection strategy based on differentiated pricing is proposed,which is introduced into the buying and selling game model through the differentiated pricing mechanism.First of all,the condition of the network is predicted by the number of requests and responses received by the vehicle,and the urgency of the task is derived from the amount of tasks generated by the vehicle and the task tolerable delay.Afterwards,Secondly,on the basis of the traditional price function,the two parameters of the busyness of the network and the urgency of the task are combined to further form a new task pricing function.Then,we analyze and prove the existence and uniqueness of the solution of the game model.Based on this theory,a distributed algorithm is used to solve the resource allocation problem of differentiated pricing.Finally,the simulation results show that,compared with existing strategy,the resource utilization of the proposed algorithm improves by at least 9.6%.2.Existing task offloading strategies in vehicular cloud computing only optimizes the time delay based on the optimal task processing time.This may cause the task processing time to exceed the task deadline or disconnect between vehicles during task transmission,resulting in task processing failure.To deal with the problem,a task offloading strategy based on optimal time delay limit is proposed.The offloading problem in vehicle cloud computing with optimal time delay is modeled as an optimization problem for maximizing task execution efficiency,in which task deadline and vehicle connection time are two factors that affect task offloading,and they will be used as constraints to further construct new optimizations model to solve the optimization problem,so as to obtain the optimal task allocation scheme.Finally,Compared with the existing task offloading strategy,the task offloading strategy proposed in this paper based on the optimal time delay limit increases the total task completion rate by 14.5%.