Research On Computing Offloading Algorithm Based On Path Predictability In Vehicular Ad Hoc Cloud

As the number of vehicles increases rapidly and the demand for intelligent vehicles increases,the computationally intensive tasks generated by resource-constrained vehicles affect the service experience of vehicular users.In the past,cloud computing was used to provide computing and storage resources for resource-constrained vehicles.The vehicle terminal uploads computing tasks to the cloud server.The process of assigning available resources to computing tasks according to user requirements and executing these by the cloud server agent is called computing offloading.The main purpose of the computing offloading is to reduce the task completion time of the local devices.However,traditional cloud computing has problems such as high execution costs,limited bandwidth resources and severe cloud loads.In the vehicular computing environment,the vehicular cloud computing can be divided into an infrastructure based vehicular cloud computing and a vehicular self-organizing cloud computing composed of heterogeneous vehicles according to whether there is infrastructure support.Due to the high dynamics of the vehicular ad hoc Cloud,the communication environment between the vehicle and the vehicle changes frequently during the computing offloading causing longer task completion time.In the computing offloading of the vehicular ad hoc cloud,how to choose the most suitable target node to offload computing is the key to ensure the efficient execution of the task.The computing offloading based on the flooding algorithm can effectively ease the influence that the target node selection has on the computing offloading.However,this algorithm will produce excessive waste of resources.Therefore,how to use existing resources to provide high-quality computing offloading services for resource-constrained vehicles under the premise of reducing resource occupancy rate is the focus of this paper.In order to solve the two problems of reducing resource occupancy rate and improving computing offloading performance,the main work of this paper is as follows:(1)Comparative analysis of the computing offloading performance on the vehicular ad hoc cloud within and without DTN support,which proves that the DTN technology can effectively optimize the task completion time;(2)comparative analysis of the minimum hops algornthm,minimum execution time algorithm and minimum completion time algorithm in the vehicular ad hoc cloud,the predictability-based and the flooding-based computing offloading algorithms to improve the computing offloading performance are proposed;(3)In order to reduce the resource occupancy rate,a partial flooding algorithm is introduced,comparing the predictability-based and the flooding-based computing offloading algorithms.The experimental results show that:(1)In terms of task waiting time and task completion time,the computing offloading performance in the DTN-based vehicular ad hoc cloud improves by 56.76%and 40.88%,compared with the DTN-supported vehicular ad hoc cloud;(2)The completion time of the predictability-based computing offloading algorithm and the partial flooding algorithm improves by 12.53%,15.20%,9.00%and 19.49%,21.95%,16.24%compared with the algorithms such as ROL,MET.MH;(3)To completion time,the partial flooding algorithm improves by 8.00%compared with the predictability-based algorithm;the predictability-based algorithm improves by 35.46%compared with the partial flooding algorithm in terms of the resource occupancy rate.