Clustering Protocol Based On Mobility Analysis In Vanets

Vehicular Ad Hoc Networks, serving as the support technology of vehicular applications, is also one of the basic technologies of Intelligent Transport Systems (ITS). Recently, it has raised attentions and investments from governments, car manufacturers, domestic and international standard organizations and research institutes. Thus a lot of communication standards and communication mechanisms have emerged in recent years. However, there are still many inter-vehicle communication issues need to be resolved. Researches on message authentication, routing and forwarding algorithms, message priority, traffic congestion control, scalability still face the challenges introduced by the special mobility characteristics in VANETs. Therefore, efficient analysis on the mobility metrics in VANETS and usage of it in vehicular communication network research is one of the important directions. Clustering, as one of the key network technologies in VANETs, constructs hierarchical topology structure, which makes the network management can be completed in a distributed manner, so the overall complexity of the vehicular network is reduced. However, VANETs with fast-moving nodes, frequently changed topology, and uneven distribution of node density, have a significant effect on the stability of the clustering in VANETs. Therefore, analyze the mobility of VANETs, and accordingly design comprehensive clustering protocol is an significantly effective way.This paper surveys the related works on mobility metrics analysis and clustering strategies, after which two main aspects of works have been done, namely mobility metrics analysis and clustering protocol designation.Firstly, the paper summarizes the advantages and disadvantages of different kinds of analysis methods of mobility metrics. Thus a three phase traffic theory based analysis scheme, which combines the macro mobility and micro mobility in VANETs, has been used to analyze the mobility features and connectivity features of different traffic flows at fnode level, node pair level, group level and network level. The results explain the possible effects of mobility on connectivity in VANETs.Secondly, based on the analysis of mobility metrics in VANETs, this paper summarizes the advantages and disadvantages of the existing clustering strategies. Thus a traffic flow states based clustering protocol is proposed. The paper also designs the traffic states detection algorithm and the dynamic transmission change algorithm are also designed for the proposed clustering protocol, to improve the performance in different traffic states and different density of nodes. In addition, the proposed protocol develops a comprehensive cluster formation strategy and a cluster maintenance strategy, and a dynamic HELLO period algorithm is proposed in order to reduce clustering overhead. Finally, the proposed clustering protocol is verified through simulation. The results show that the proposed clustering protocol outperforms the existing clustering protocols in cluster life time, cluster residence time, Re-affiliation Rates and number of clusters.