面向车载自组织网络路由的轨迹预测算法

发布时间：2019-06-17 15:47

【摘要】：在车载自组织网络(vehicular ad hoc network,VANET)(也称车联网)中,基于地理位置的路由协议能够较好地适应网络拓扑的动态性变化和链路质量的不稳定性.由于位置信息需要在邻居节点间采用信标分组进行交互,信标分组间隔内的转发决策可能因车辆节点位置的移动而不准确,需要进行位置预测来修正车辆节点的位置.已有的位置预测算法存在普适性差或预测误差大的问题.针对上述问题,提出了一种新的预测算法,首次通过测量得到车辆加速度服从正态分布的结论,利用线性回归进行预测,并采用反馈机制进行结果修正.利用真实车辆轨迹进行测试,新的预测算法的预测精度大为提高.然后,提出了一种新的基于位置的即时路由协议.在该协议中,发送节点利用邻居节点位置和目的节点位置计算出转发下一跳.将新的位置预测算法加入到即时路由协议中,实时预测和更新车辆的位置.利用SUMO软件生成了基于真实地图道路轨迹的车辆运动模型,结合NS3网络仿真平台进行了仿真实验.实验结果表明:采用新的预测算法后,相比传统的GPSR协议和不带预测的即时路由协议,新方法的收包率提高、延迟下降,并且协议开销显著降低.
[Abstract]:In the vehicle-mounted ad hoc network (VANET) (also known as the vehicle networking), the routing protocol based on the geographical location can better adapt to the dynamic change of the network topology and the instability of the link quality. Because the location information needs to be interacted with the beacon packet between the neighbor nodes, the forwarding decision within the beacon packet interval may be inaccurate due to the movement of the vehicle node position, and the position prediction is required to correct the position of the vehicle node. The existing position prediction algorithm has the problem of large universality or large prediction error. In view of the above-mentioned problems, a new prediction algorithm is proposed, which is based on the conclusion that the vehicle acceleration obeys the normal distribution, the linear regression is used for the prediction, and the feedback mechanism is used for the result correction. The real vehicle trajectory is used for testing, and the prediction accuracy of the new prediction algorithm is greatly improved. Then, a new location-based instant routing protocol is proposed. In this protocol, the sending node calculates the forwarding next hop using the neighbor node location and the destination node location. The new position prediction algorithm is added into the instant routing protocol, and the position of the vehicle is predicted and updated in real time. The vehicle motion model based on the real map road trajectory is generated by the SUMO software, and the simulation experiment is carried out in combination with the NS3 network simulation platform. The experimental results show that compared with the traditional GPSR protocol and the non-predicted real-time routing protocol, the packet rate of the new method is improved, the delay is reduced, and the protocol overhead is significantly reduced.
【作者单位】：清华大学计算机科学与技术系;
【基金】：华为公司创新研究计划项目国家自然科学基金项目(61602271,61373143,61432009) 中国博士后科学基金项目(2016M591182)~~
【分类号】：TN929.5

【相似文献】