Research On Vehicle Delay-tolerant Network Routing Algorithm Based On Bayesian Classifier

Vehicle Delay tolerant network(VDTN)is a wireless mobile network,where constant end-to-end connections between source nodes and destination nodes may not exist at any given time.Due to intermittent contact and disconnections,the nodes adopt the “store-carry-and-forward” mechanism to deliver messages.Therefore,traditional end-to-end routing protocols are not applicable.Researchers need to study more effective routing protocols according to the characteristics of VDTN,and make use of multi-hop cooperation between nodes and the opportunity of node encounter.Such routing protocols must be flexible and be able to adapt to rapidly changing network topology.Firstly,the related theories and key technologies of VDTN have been described in this thesis.On this basis,we focus on the repetitive movement patterns and the high-order dependencies between node attributes.On this basis,considering the repetitive movement patterns of nodes in VDTN,a routing algorithm based on Semi-naive Bayesian classifier(SBC)is proposed,which adopts an approach for estimating the delivery probability by utilizing the network parameters(e.g.spatial and temporal information at the time of packet forwarding)that concern the movement patterns of DTN nodes.SBC introduces the “One-Dependent Estimator” where the basic idea is to assume that all attributes depend on the same attribute.This algorithm focuses on the dependencies among attributes to provide a more accurate model for the classification of nodes.Moreover,this thesis considers the complexity of node movement patterns in VDTN,expands the attribute set of the node,and introduces the Bayesian network classifier to improve the generalization performance by studying the high-order dependencies between attributes.Considering the multi-stage characteristics of the node mobility patterns,the Multi-period Bayesian Network(MBN)is proposed to build multiple prediction models,which intends to predict the regular movement patterns of nodes in the real world.Additionally,a novel classification method called Dynamic Multiple-Level Classification(DMLC),is proposed where nodes are classified into multiple levels according to the dynamic parameters.In addition,in order to capture the dynamic characteristics of the network,we define delivery bonus as an attribute of each node,which is updated dynamically according to the time,encounter frequency and message delivery results.The delivery bonus has a similar refresh mechanism with the pheromone of ant colony algorithm,which is featured as positive feedback and self-adapting optimization.Due to the more effective information of node movement and the high-order dependencies between attributes,the node can make better routing decisions in the process of delivering messages.The simulation results show that the routing algorithm based on Semi-naive Bayesian Classifier can significantly enhance the message delivery ratio and reduce the overhead ratio.The research in this thesis not only can widely apply to the real world,but also can provide ideas for the study on routing protocols in VDTN,which has a good theory value and application prospect.