【作者】 王辉；

【导师】 彭鑫；

【作者基本信息】 湖南理工学院 ， 信息与通信工程， 2021， 硕士

【摘要】 21世纪以来,车辆数目剧增,车载多媒体、巡航系统和自动驾驶等新型的车载应用大量涌现,这些应用在使用过程产生了大量数据,这对车联网的计算能力与处理时延提出了更高的要求。而在车联网中,车辆的计算能力有限,不足以支撑这些应用的计算处理。在此情景下,基于云计算的车联网应运而生,由于云计算部署位置距离车辆相对较远,只能解决上述要求中计算问题,无法满足时延要求,导致车辆服务质量(Quality of Service,Qo S)未得到明显改善。而移动边缘计算(Mobile Edge Computing,MEC)服务器常常部署在路边单元(Road Side Unit,RSU)或5G(5th-Generation,5G)基站上,可以很好地弥补云计算时延高的不足。在基于MEC的车联网中,当车辆卸载自身计算任务时,车辆自身的任务量大小、任务优先级、路边基础设施的可用资源和路边基础设施是否损坏等因素都会影响车辆计算卸载对象选择,选择不当的卸载对象,会存在车辆能耗、成本和时延增加等问题。针对卸载对象选择问题,本文提出了以下两种策略:(1)在城市交通中,路边基础设施一般而言是完好的,车辆可以将自身计算任务卸载路边基础设施。本文提出了一种基于车辆任务优先级的多指标下的MEC服务器选择策略。在问题建模时,本文策略基于802.11P的通信协议,同时引入任务优先级,并在车辆排队等待卸载计算任务至RSU过程中,采用最优拍卖机制,使车辆收益与RSU收益之间实现纳什均衡,使车辆能够选择最合适的卸载对象。仿真结果表明,该方案能在保障车辆Qo S的基础上,降低任务卸载的总费用、总时延与总能耗,满足多重性能指标。(2)当车辆行驶在山区道路中,因自然灾害损坏路边基础设施并且人员被困时,用户无法通过路边基础设施与外界通信,而无人机(Unmanned Aerial Vehicle,UAV)由于灵活的部署能力与机动性能够临时充当MEC服务器的角色。本文提出了一种基于UAV的加权能耗最小化的计算速率最大化方案。在该方案中,UAV拥有无线供电能力,用户消耗的能量均来自设备采集到的能量。通过联合优化UAV轨迹、CPU频率与发射功率对用户与UAV的能耗加权和进行优化,并在给定轨迹下,求解出部分卸载模式与二进制模式下用户设备最佳的CPU频率和发射功率,并求得二进制模式下,用户计算任务最佳卸载策略。仿真表明,在本文所提方案下求解所得轨迹的资源分配方案更优,收敛更快。更多还原

【Abstract】 Since the beginning of the 21 st century,the number of vehicles has increased dramatically,and a large number of new on-board applications such as on-board multimedia,cruise system and automatic driving have emerged.These applications produce a large amount of data in the use process,which puts forward higher requirements for the computing power and processing delay of the Internet of vehicles.In the Internet of vehicles,the computing power of vehicles is limited,which is not enough to support the computing processing of these applications.In this scenario,the Internet of vehicles based on cloud computing arises at the historic moment.Because the cloud computing deployment location is relatively far away from the vehicle,it can only solve the computing problems in the above requirements,and can’t meet the delay requirements,resulting in the vehicle quality of service(Qo S)has not been significantly improved.The mobile edge computing(MEC)server is often deployed on the roadside unit(RSU)or 5g(5th generation,5g)base station,which can make up for the high delay of cloud computing.In the Internet of vehicles based on MEC,when the vehicle unloads its own computing tasks,the vehicle’s own task size,task priority,available resources of roadside infrastructure and whether the roadside infrastructure is damaged or not will affect the selection of vehicle computing unloading objects.Improper selection of unloading objects will lead to the increase of vehicle energy consumption,cost and delay.To solve the problem of unloading object selection,this paper proposes the following two strategies.(1)In urban traffic,the roadside infrastructure is generally in good condition,and vehicles can unload their own computing tasks from the roadside infrastructure.In this paper,a MEC server selection strategy based on vehicle task priority is proposed.In the problem modeling,this strategy is based on 802.11 p communication protocol,and introduces task priority.In the process of vehicle queuing to unload computing tasks to RSU,the optimal auction mechanism is adopted to achieve Nash equilibrium between vehicle revenue and RSU revenue,so that vehicles can choose the most suitable unloading object.Simulation results show that the scheme can reduce the total cost,total delay and total energy consumption of task unloading on the basis of ensuring vehicle Qo S,and meet multiple performance indicators.(2)When vehicles are driving on mountain roads,the roadside infrastructure is damaged due to natural disasters and people are trapped,users can not communicate with the outside world through the roadside infrastructure,and unmanned aerial vehicle(UAV)can temporarily act as MEC server due to its flexible deployment ability and mobility.This paper proposes a computing rate maximization scheme based on UAV weighted energy minimization.In this scheme,UAV has the ability of wireless power supply,and the energy consumed by users comes from the energy collected by the device.Through the joint optimization of UAV trajectory,CPU frequency and transmit power,the weighted sum of user and UAV energy consumption is optimized.Under the given trajectory,the optimal CPU frequency and transmit power of user equipment in partial unload mode and binary mode are solved,and the optimal unload strategy of user computing task in binary mode is obtained.Simulation results show that the resource allocation scheme is better and converges faster.更多还原