车辆边缘计算环境下任务卸载研究综述

计算密集和延迟敏感型车辆应用的出现对车辆设备有限的计算能力提出了严峻的挑战,将任务卸载到传统的云平台会有较大的传输延迟,而移动边缘计算专注于将计算资源转移到网络的边缘,为移动设备提供高性能、低延迟的服务,因此可作为处理计算密集和延迟敏感的任务的一种有效方法.同时,鉴于城市地区拥有大量智能网联车辆,将闲置的车辆计算资源充分利用起来可以提供巨大的资源和价值,因此在车联网场景下,结合移动边缘计算产生了新的计算模式——车辆边缘计算.近年来,智能网联车辆数量的增长和新兴车辆应用的出现促进了对车辆边缘计算环境下任务卸载的研究,本文对现有车辆边缘计算环境下任务卸载研究进展进行综述,首先,从计算模型、任务模型和通信模型三个方面对系统模型进行梳理、比较和分析.然后介绍了最小化卸载延迟、最小化能量消耗和应用结果质量三种常见的优化目标,并按照集中式和分布式两种不同的决策方式对现有的研究进行了详细的归类和比较.此外,本文还介绍了几种常用的实验工具,包括SUMO、Veins和VeinsLTE.最后,本文围绕卸载决策算法复杂度、安全与隐私保护和车辆移动性等方面对车辆边缘计算任务卸载目前面临的挑战进行了总结,并展望了车辆边缘计算环境下任务卸载未来的发展方向与前景.

A Survey on Task Offloading Research in Vehicular Edge Computing

The emergence of computation intensive and delay sensitive vehicle applications poses a severe challenge to the limited computing capacity of vehicle equipment.Offloading tasks to traditional cloud platforms have large transmission delays,and the cost of upgrading on-board computers is huge,so these two methods have some disadvantages in dealing with computation intensive and delay sensitive tasks.Mobile edge computing is a new computing paradigm,which focuses on transferring computing resources to the edge of the network,providing high performance,high reliability and low latency services for mobile devices.Therefore,it will be a more effective way to process computation intensive and delay sensitive tasks.Meanwhile,vehicles can act as both service requesters and service providers.In view of the large number of intelligent networked vehicles in urban areas, making full use of idle vehicle computing resources can provide huge resources and value.Therefore,combined with the mobile edge computing,a new computing paradigm is generated in the Internet of Vehicles scenario,called vehicular edge computing(VEC).In recent years,the increase of the number of intelligent networked vehicles and the emergence of emerging vehicle applications have promoted the research on task offloading in vehicular edge computing.However,there is no detailed summary and analysis of the problems related to task offloading in VEC at present.This paper summarizes the research progress of task offloading in the existing vehicular edge computing.Firstly,the VEC system model is summarized,compared and analyzed,including computing model,task model and communication model.Specifically,the VEC computing model consists of a three-layer cloud structure of remote cloud(RC),edge cloud(EC),and vehicular cloud(VC),each of which has its own advantages in different aspects.The VEC task model is divided into critical applications(CAs),high-priority applications(HPAs) and low-priority applications(LPAs) according to the degree of application criticality to vehicles,and the dependencies between tasks are summarized.Several main communication protocols for task offloading in the Internet of Vehicles are introduced in the VEC communication model,and the influence of vehicle mobility on communication is analyzed.Secondly,we summarize three common optimization objectives in the task offloading of VEC from existing studies,namely,minimizing offloading delay,minimizing energy consumption and application quality of results.Thirdly,we summarize some models and methods of VEC task offloading,such as semi-Markov decision process,game theory,reinforcement learning,heuristic algorithm and contract theory.At the same time,the existing researches are classified according to centralizing and distributing two different decision-making methods,and the characteristics of these two decision-making methods are compared.Then,we describe several commonly used experiment tools,including SUMO,Veins and VeinsLTE,which can make simulation experiments more realistic and credible.In addition,the current challenges of task offloading in vehicular edge computing are summarized in section 6,including offload decision algorithm complexity,security and privacy protection and vehicle mobility.In view of the above problems,we have proposed some possible future research directions and development prospects.Finally,we summarize the whole paper in section 7.