【作者】 张林涛；

【导师】 唐余亮；

【作者基本信息】 厦门大学 ， 通信与信息系统， 2022， 硕士

【摘要】 依托于5G网络技术的蜂窝车联网(Cellular-Vehicle-to-Everything,C-V2X)技术不仅可以满足高速率、广域覆盖的需求,还将满足未来车载应用更为严苛的高可靠、低时延需求。为了解决终端车辆资源受限问题,将移动边缘计算(Mobile Edge Computing,MEC)整合进C-V2X形成基于移动车辆边缘计算(Vehicle Edge Computing,VEC)的蜂窝车联网来为用户提供低时延服务。考虑到VEC所具有的计算、存储和通信资源的限制,需通过多域资源的协同优化来提高资源利用率、提高系统吞吐量、实现负载均衡和降低系统时延。本文对VEC中多域资源分配和优化进行研究,内容如下:首先,针对C-V2X有限频率资源难以有效支撑VEC中大规模车辆用户的接入问题,将缓存辅助的非正交多址接入(Non-Orthogonal Multiple Access,NOMA)技术引入到VEC下行数据传输中。同时,为了提高所有车辆用户的服务质量,并且保证服务公平性,以最大化最小车辆平均中断数据率为优化目标,提出了功率分配和串行干扰消除(Successive Interference Cancellation,SIC)排序的联合优化方案。考虑到非完美信道估计的实际情况,借助马尔可夫不等式将涉及概率约束的问题推导为非概率约束问题,然后将问题分解为功率分配和SIC排序两个子问题。提出了基于二分法和元贪婪的功率分配和SIC排序算法。仿真结果显示,本文提出的方案在传输速率、算法时间复杂度等方面有了较大提高。其次,针对C-V2X中多边缘服务器VEC系统中具有时间和数据双重依赖关系的任务调度、缓存决策和负载均衡问题设计分层资源管理框架,提出了最小化系统时延的优化问题,将问题分解为所提分层结构框架中的单VEC服务器内资源管理问题和多VEC服务器间资源协调问题,设计相应算法求解。在VEC服务器内的资源管理中以降低平均任务完成时延为目标,提出了基于最小最迟开始时间的任务调度算法和基于动态规划的缓存决策算法;在VEC服务器间资源协调中,以降低系统时延为目标,提出了基于联盟博弈的负载均衡算法。仿真结果显示,本文提出的方法在任务时延、失败率以及资源利用率等方面性能都有较大提高。更多还原

【Abstract】 Cellular-Vehicle-to-Everything(C-V2X)technology,which relies on 5G network technology,will not only meet the needs of high-speed,wide-area coverage,but also meet the more stringent high reliability and latency-sensitive needs of future in-vehicle applications.Integrate Mobile Edge Computing(MEC)into C-V2X to form a cellular vehicle networking based on Vehicle Edge Computing(VEC)to provide users with lowlatency services.Considering the limitations of computing,storage and communication resources of VEC,it is necessary to improve resource utilization,improve system throughput,achieve load balancing and reduce system latency through collaborative optimization of multi-domain resources.In this paper,we study the allocation and optimization of multi-domain resources in VEC,and the research content is as follows:Firstly,in view of the difficulty of C-V2X limited frequency resources to effectively support the access problem of large-scale vehicle users in VEC,the cache-assisted NonOrthogonal Multiple Access(NOMA)technology is introduced into the VEC downlink data transmission.To improve the quality of service(QoS)of all vehicles and guarantee fairness,we formulate the power allocation and SIC ordering selection(PASO)problem to maximize the minimum achieved average outage data rate for vehicles.Considering the imperfect channel estimation,markov’s inequality is used to transform the probabilistic constraint problem into a non-probability problem,and then the problem is decomposed into two sub-problems of power allocation and SIC ordering,and a power allocation and SIC ordering algorithm based on bisection method and greedy-meta scheduling is designed.Simulation shows that the proposed scheme has greatly reduced the algorithm time complexity and improved the transmission rate.Secondly,for the computing tasks with complex dependencies between subtasks in C-V2X,a VEC resource management framework with hierarchical structure is proposed and corresponding algorithms are designed.Aiming at the problems of task scheduling,cache decision-making and load balancing with dual dependence on time and data in the multi-edge server VEC system,the optimization problem of minimizing system latency is proposed.The problem is decomposed into the resource management problem of single VEC server and the resource coordination problem between multiple VEC servers in the proposed hierarchical framework,and the corresponding algorithm is designed to solve it.In the resource management of VEC server,a task scheduling algorithm based on the minimum latest start time and a cache decision algorithm based on dynamic programming are designed to reduce the average task completion time,and in the resource coordination between VEC servers,a load balancing algorithm based on coalition game is designed with the goal of reducing system latency.Simulation results show that the proposed resource management framework and algorithm make great improvements in task latency,failure rate and resource utilization.更多还原