The Design And Implementation Of VANET Cloud Service System Based On Hadoop

Vehicular Ad hoc Networks(VANETs) is an important part of Intelligent Transport System(ITS). The research in VANETs is helpful to improve the transportation system, supply chain management and logistics. As vehicle terminals becoming more powerful in data processing and communication, one simple application of the service can produce and gather a great mount of data. Meanwhile, data types become more complicated and the processing load of the vehicular system increasingly heavier. How to solve the storage and processing of the massive complex data in VANETs and how to ensure the communication propagation of the vehicle effectively are always key issues in the research of VANETs. Nowadays cloud computing and big data technologies have attracted considerable attention and will be applied in different research areas. This dissertation mainly focuses on several essential problems in VANETs. We set up the VANET cloud based big data platform using Hadoop. After that we put the vehicular application data storage and the computing all on the cloud, and use the distributed framework MapReduce to process batch data for routing the shortest path. Finally, we release the big data platform as a whole service to meet the needs of significantly enhancing the vehicular users’experience.This dissertation consists of the following major aspects:(1) We carry out an investigation about the big data problems in VANETs. This dissertation analyzes the vehicular information data flow, distribution, dynamic and uncertainty characteristics, so as to map the "4V" characteristics of big data and verify that most of the problems in VANET, and hence, can be treated as big data problems to explore new ideas for solution.(2) We set up a vehicular cloud big data processing platform. In this dissertation, we use the open-source Hadoop as the basic platform to build vehicular cloud big data processing platform based on Hadoop. It is compatible with offline, near real-time and real-time environment under different data size, to solve the complicated and massive data storage and processing problem in VANETs.(3) We design and implement a distributed parallel routing algorithm. The routing protocol ensures that the effective route connection is established between the vehicle nodes, and the effectiveness of the route connection is adjusted with the change of the network topology. According to the characteristics of VANETs, this dissertation designs and implements a distributed Dijkstra algorithm to find the shortest path for information transferring, and evaluates the performance of the processing time in different number of Hadoop clusters.