车载容迟网络下基于节点传输能力的服务分发协议

摘要

The node-delivery-capability-based service dissemination (NDCSD) protocol is proposed to solve the degrading performance issue for data transmission in vehicular delay tolerant networking (VDTN) due to the frequently intermitted connections.In the NDCSD protocol,the node delivery capability is computed by the integration of the throughput function within the active connection time.The wavelet-based neural network is used to train the sampling data to obtain the approximation of the throughput function.Then,through taking the node delivery capability as the metric,the service data distribution protocol is designed by using the Dijkstra algorithm,and the NDCSD protocol is implemented on the opportunity networks emulator.The experimental results show that compared with other classical protocols,the NDCSD protocol can significantly improve the successful rate of service delivery and reduce the data transmission delay,the delay jitter and the packet loss.The combination of the connection time and the throughput of nodes can improve the service delivery capability and implement the reliable transmission.%为了解决车载容迟网络(VDTN)中节点间歇性连接导致的传输性能下降问题,提出了基于节点传输能力的服务分发协议(NDCSD).在NDCSD协议设计中,通过对节点有效连接时间估计内的吞吐量函数积分得到节点传输能力,采用小波神经网络对样本数据进行学习的方法获得吞吐量函数估计.然后,以节点传输能力为度量,采用Dijkstra算法设计服务数据分发协议,并在随机网络仿真器上实现了该协议.结果表明,与同类协议相比,NDCSD协议能够提高VDTN环境下服务数据的投递成功率,降低服务数据的传输时延、时延抖动及丢包率,说明综合考虑连接时间和吞吐量的方法能够改善VDTN下的服务传输能力,实现可靠传输.