摘要
浮动车数据(floating car data, FCD)技术是大规模城市路网交通流实时采集的有效方法.城市交通的动态诱导和控制需要对海量FCD进行快速处理.鉴于此,提出了FCD并行计算的动态任务调度方法.针对FCD数据包计算时间的不确定性和动态性,根据计算节点的处理能力进行数据包的动态分割,在处理过程中,采用动态任务分配策略以实现计算节点的同步.该方法在龙芯国产大数据一体机平台上进行了实现,并采用现场FCD数据进行了实验验证,结果表明,该方法较轮询和Min-Min调度算法,显著地提高了并行处理的性能.
FCD(floating car data)technique is new way of collecting real-time traffic flow from large-scale urban networks.It is necessary to implement rapid processing of FCD big data for the dynamic guidance and control of urban traffic.A dynamic task scheduling algorithm is proposed for parallel computation of FCD.To address the uncertainty and dynamics of FCD package processing,FCD packages are partitioned dynamically.The load balance among computing nodes can be achieved using the dynamic task allocation strategy.The algorithm is developed on LoongSon big data integrated machine platform and evaluated using field FCD.The experimental results indicate that the proposed algorithm has significantly higher parallel processing performances compared to the polling scheduling algorithm and Min-Min scheduling algorithm.
作者
陈锋
张智
李琴剑
陈宇强
陈国良
CHEN Feng;ZHANG Zhi;LI Qinjian;CHEN Yuqiang;CHEN Guoliang(Department of Automation,University of Science and Technology of China,Hefei 230027,China;Anhui LoongSon Science and Technology Co.,Ltd,Hefei 230088,China;School of Computer Science and technology,University of Science and Technology of China,Hefei 230027,China)
基金
国家重点研发(973)计划(2017YFC0840206)
安徽省重大科技专项(17030901007)资助
关键词
浮动车数据
大数据
并行计算
动态任务划分
动态任务调度
floating car data
big data
parallel computing
dynamic task partition
dynamic task scheduling
