Research On Multitask Allocation Strategy Based On Task Fragmentation In Autonomous Vehicular Cloud

Vehicles traveling on highways,when there is a large computational task to be performed,a single vehicle cannot complete it due to its own computing power and storage performance limitations,and needs to allocate it to surrounding vehicles for execution.The vehicles traveling on the highway can interconnect together to form a vehicular ad hoc networks.The autonomous vehicular cloud combines the vehicular ad hoc networks with the cloud.The vehicle in the autonomous vehicular cloud can allocate its task to one or more surrounding vehicles for execution.In the scenario of the autonomous vehicular cloud in the highway,there may be more than one vehicle with such task execution requirements for a period of time.Because the highway road conditions are complex,the vehicle nodes move faster,the communication between vehicles is intermittent,and the node topology changes frequently,how to allocate such large computational tasks initiated by multiple task-initiating vehicle nodes to the surrounding nodes in the autonomous vehicular cloud highway scenario,becomes an urgent problem to be solved.In this thesis,by considering whether multi-tasks in the autonomous vehicular cloud highway scenario need to be fragmented and how to be fragmented,a multi-task fragmentation and allocation strategy is proposed.By considering the communicable duration and computing power between vehicle nodes,the executable task slice size of the surrounding nodes is determined.The task execution nodes are selected,and the large computational tasks are divided into pieces.Under the condition of considering fault tolerance,channel interference and queue waiting delay,multiple large computational tasks are implemented between multiple task execution vehicle nodes.The results of OPNET simulation experiments show that the multi-task allocation strategy based on executable task fragments is compared with the multi-task allocation strategy of non-fragmented randomly selected task execution nodes and the multi-task allocation strategy of equal-partitioning fragments.Comparing with these two comparison experiments,the multi-task completion rate of the strategy in this thesis is increased by 17.83% and 54.67% respectively,and the multi-task completion delay is reduced by 53.67% and 25.82% respectively.