摘要
针对车载自组织网络中移动车辆间的多跳数据传输,结合城市车辆移动特征,通过改进现有的BubbleRap路由机制,提出一种基于分布式学习的数据转发机制(DFDL)。该机制基于存储-携带-转发的消息传输模式,利用移动车辆间相遇时间间隔和相遇频率确定车辆的社区标签,并根据车辆运动的移动熵计算节点运动中心度。在转发过程中DFDL机制通过综合判断相遇车辆的社区标签以及运动中心度,为数据包消息选择合适的中继转发节点,实现远距离移动车辆间的多跳消息转发。实验结果表明,对于城市车辆移动场景,与BubbleRap路由机制相比,DFDL机制的消息投递率及平均投递延时性能明显提升。
Aiming at the problem of multi-hop data transmission among mobile vehicles in Vehicular Ad Hoc Network( VANET),combining the moving features of vehicle in cities,this paper proposes a Data Forwarding mechanism based on Distributed Learning( DFDL) by improving the existing BubbleRap routing mechanisms. Based on the "store-carry-and forward"message transmission mode,DFDL determines the vehicles' community tags by using the time interval and frequency of enwunter,and calculates the centrality of nodes by the mobile entropy of vehicle movement. During data forwarding,DFDL selects appropriate relay nodes for data packet messages by comprehensively judging the community tags and centrality of nodes. This mechanism achieves multi-hop message forwarding among moving vehicles in long distance. Experimental results showthat,for moving vehicles in cities,DFDL significantly improves the message delivery ratio and the average delivery delay performance than the BubbleRap routing mechanism.
出处
《计算机工程》
CAS
CSCD
北大核心
2016年第3期1-6,共6页
Computer Engineering
基金
国家博士后科学基金资助面上项目(2014M560867)
北京市博士后工作经费基金资助项目
关键词
车载自组织网络
社区检测
移动延迟容忍网络
运动模型
机会传输
Vehicular Ad Hoc Network(VANET)
community detection
mobile Delay Tolerant Network(DTN)
mobile model
opportunistic transmission