车载云计算系统中的资源管理优化研究  

Research on Resource Management and Optimization in Vehicle Cloud Computing System

作　　者：张文萍 陈桂芬[1] 刘可欣 ZHANG Wen-ping;CHEN Gui-fen;LIU Ke-xin(School of Electronics and Information Engineering,Changchun University of Science and Technology,Changchun 130022)

机构地区：[1]长春理工大学电子信息工程学院,长春130022

出　　处：《长春理工大学学报（自然科学版）》2020年第6期102-112,共11页Journal of Changchun University of Science and Technology(Natural Science Edition)

基　　金：吉林省科技厅项目(20190302103G X)。

摘　　要：车联网是物联网(Internet of Things,IOT)技术在智能交通领域的典型应用,研究车联网关键技术,可以高效促进我国交通系统建设。车载云计算(Vehicular Cloud Computing,VCC)作为实现智能交通的关键技术之一,在降低功率和时间的消耗,提高车辆总体资源利用率和系统长期收益等方面具有至关重要的作用。针对车辆自身资源受限以及将任务卸载到中心云将导致较高通信成本的情况,提出在车载云之间引入服务迁移的机制,同时将路边单元(Road Side Unit,RSU)和车辆异构性考虑进VCC系统中,基于半马尔科夫决策过程(Semi-Markov Decision Processes,SMDP)建立了VCC系统模型,最后应用值迭代算法求解,来寻找VCC资源分配的最优策略。仿真结果展示了车辆异构性对资源分配的影响,同时表明了SMDP资源管理方案的优越性,SMDP相比于贪婪算法(Greedy Algorithm,GA)和模拟退火算法(Simulated Annealing,SA)这两个传统算法,系统长期收益分别提高了10%和3%左右。Internet of vehicles is a typical form of Internet of Things(IOT)technology applied in the field of intelligent transportation.Research on key technologies of Internet of vehicles is of great significance for the construction of Chinese transportation system and national economic development.As one of the key technologies to realize intelligent transportation,Vehicular Cloud Computing(VCC)plays a vital role in reducing power and time consumption,improving overall resource utilization rate of vehicles and long-term benefits of the system.Against its own limited resources and the situation that offloading to the central cloud will lead to the high communication cost,proposes between VCCs into service migration mechanism,meanwhile,the Road Side Unit(RSU)and vehicle heterogeneity are considered into the VCC system,based on Semi-Markov Decision Process(SMDP)VCC system model is set up.Finally,we apply value iteration algorithm to find the VCC optimal strategy of resource allocation.The simulation results show the vehicle heterogeneity influence on the allocation of resources and the superiority of SMDP resource management scheme.Compared with the traditional Greedy Algorithm(Greedy Algorithm,GA)and Simulated Annealing Algorithm(Simulated Annealing,SA),SMDP improves the system long-term return by about 10%and 3%respectively.

关 键 词：车载云计算 车辆异构性 半马尔科夫决策过程 服务迁移 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]