摘要
基于车载自组织网络的特性,提出一种借助于梯度场的方法,并将神经网络应用于车载自组织网络进行下一跳节点选择的路由算法,以达到快速准确地传递数据包的目的。该算法利用节点的位置、速度等信息,计算节点的梯度值,并利用神经网络根据不同路段的条件调整梯度计算中各个参量的优先级,选择梯度值最大的节点作为下一跳节点。仿真结果表明,与城市场景下的贪婪边界无状态路由(GPSR,Greedy Perimeter Stateless Routing)和无线自组网按需平面距离矢量路由(AODV,Ad hoc On-Demand Distance Vector Routing)相比,基于神经网络和节点梯度的路由协议(NN-NGR,Neural Network and Node Gradient Routing)在数据包丢包率、数据包端到端平均时延方面具有较好的性能。
Based on the characteristics of the VANET, an approach was proposed to choose a next node to receive the packet with the help of gradient field method and neural network, thus to realize rapid and accurate dissemination of the packets. Based on the location and driving speed information of each vehicle, node gradient was calculated. Neural network was used to adjust the priority of parameters according to the road conditions when calculating node gradient, then choose the vehicle node whose gradient is biggest. Simulation results show that NN-NGR has better performance than GPSR and AODV in urban scenario in terms of packet drop ratio and average end-to-end delay.
出处
《电脑与信息技术》
2016年第4期44-48,共5页
Computer and Information Technology
