基于 VANETs 下的决策树多副本机会路由协议

在车载自组织网络（vehicular Ad hoc networks，VANETs）中，当节点缓存和消息副本数目被限制的情况下，如何合理地选择车载网络的路由节点是实现VANETs高效转发和投递的关键问题。为此提出了一种基于学习方法的决策树理论的多副本VANETs机会路由协议（D-Tree）。D-Tree将VANETs中节点间的传输和连接因素看做多个属性的集合，并与决策树方法得到一个消息转发规则，同时结合多副本路由与机会路由的“存储─携带─转发”优势进行消息投递。真实数据集上的实验结果表明，在场景密集的情况下，D-Tree相比于Bubble和S＆W路由算法投递成功率提高了近10％，同时在投递延迟等方面也具有明显优势。

Multiple copies of VANETs based on decision-tree for opportunistic routing protocol

In order to effectively forward and deliver the messages in the Vehicle Ad Hoc Networks (VANETs), the selection of routing node is the key issue in the Vehicle Networks in the case of fixed node buffer and limited number of message copy. Therefore, in this paper, we propose an effective opportunistic routing protocol for the multiple copies of VANETs based on the decision tree theory (D-Tree). The proposed scheme takes the intermediate node''s transmission and connection as the set of multiple attributes, and then we can obtain a rule of message forwarding based on the attributes and D-Tree. Furthermore, we can combining the advantages of multiple copies routing scheme and adopting the storage-carry-forwarding of opportunistic routing to deliver the messages. Simulation results confirm that our proposed D-Tree network model, decision rule procedure and calculating method improve the success rate of delivery compared with the conventional Bubble routing algorithm by 10% under the case of density environment. The proposed scheme also demonstrates better performance in delivery delay.