Massive Floating Car Map Matching Base On Cloud Platform

Traffic congestion has become a bottleneck of urban development and badly affects people’s living quality of daily life and happiness. In order to improve traffic service in the city, most developed countries are developing Intelligent Transportation System. As an intelligent transportation data collection system, floating car technology provides basic data for Intelligent Transportation System. Map Matching is an important component of floating car system which can convert GPS data in floating car system into traffic information.This paper studies Map Matching technology applied in massive floating car data.Compared with real-time floating car GPS data, massive historical data of floating car is characterized by large magnitude and unstable sampling interval. With the increase of floating car data magnitude, stand-alone system computing is unable to achieve fast map matching of GB or more floating car data.Cloud computing has develped into a new comprehensive application in response to magnanimity data processing. Supporting technologies(Distributed file System, MapReduee, NoSQL database, etc) of cloud computing have an inestimable advantage over conventional technology in magnanimity data processing. By importing floating car data into cloud computing environment, we can use these supporting technologies to reduce the difficulty of magnanimity data processing, save cost and improve processing efficiency.This paper optimizes map matching of floating car data in algorithm and platform. 1. In platform, we choose open source distributed architecture- Hadoop as cluster system instead of stand-alone system. Hadoop Distributed File System makes it possible to achieve distributed storage of massive floating car data and MapReduce programming model enables to realize parallel map matching of massive floating car data. 2. In map matching algorithm, This paper uses a quadtree road filtering algorithm, compared to ordinary grid road filter algorithm, quadtree structure supplies a better solution to the uneven distribution of road network density and optimizes candidate matching section.In map matching of massive floating car data experiment based on Hadoop, our algorithm excels in accuracy and speed, especially a notable improvment in speed compared with stand-alone system. Therefore, this paper has not only innovative significance in theory, but also great practical application value.