基于关联规则挖掘的VANET环境下车辆路径预测

Vehicle route prediction using association rule based mining in VANET

通过关联分析法构建相关模式数据挖掘算法,实用性强,但对具有时空动态性的车载自组织网络VANET进行车辆路径预测仍具有局限性,根据VANET环境下车辆数据特点,提出一种基于强关联规则频繁模式的车辆路径的挖掘优化算法.序列模式以有序的方式描述了事件的发生,在VANET环境中,车辆路径序列表达为车辆从起点到目的地行程中经过的路段顺序列表.首先提出基于安全认证的车辆数据采集方案,将收集的数据存储于路边单元RSU中,然后采用频繁模式数据挖掘方法分析了收集到的车辆路径,确定了某一区域内车辆选择的常见路径和频繁路径,最后本文算法与中间节点选择算法INSA进行了比较,对车辆在异常情况下选择的路径进行了评估与预测,分析结果表明:本文算法在网络通信开销、吞吐量和数据包传送率方面优于INSA算法,具有较好的置信度,有助于用户减少紧急情况下和正常情况下的等待时间.

Route patterns prediction in Vehicular Ad-hoc Network(VANET)has been receiving increasing attention due to the enabling of on-demand,intelligent traffic analysis and real time responses to traffic issues.Sequential patterns describe the occurrence of events in a timely and ordered fashion.In VANET,these events are defined as an ordered list of road segments traversed by vehicles from a starting point to final intended destination.The data collection scheme based on secure authentication to collect the data from the vehicles is proposed in the paper.The collected data is stored at Road Side Units(RSU).From these collected routs,the common and frequent routes opted by the vehicles in a certain region are determined by using data mining approach.The proposed data collection scheme is compared with INSA and the results show that the proposed scheme has better performances than INSA in communication overhead,throughput and packet delivery ratio.An estimation model is used to decide the next route opted by the vehicles in unusual situations.The proposed scheme will help users in reducing the waiting time in emergency situations and normal conditions.