Research And Implementation Of Distributed Parallel Map-matching Algorithm Based On Hadoop

With the rapid development of intelligent transportation system(ITS), geographical information technology, satellite-positioning technology and modern communication technology have played a huge role in solving the intelligent transportation in cities. Floating car data, which is an important part of ITS, is a new mode of city transportation planning and getting traffic information. Map-matching technology is one of the most important contents of the data processing on floating car data. Only judging the vehicle in which way, can the GPS data be effectively converted to road traffic condition information.Cloud computing,which can deal with large-scale floating car data quickly and effectively, is a distributed computing. In this calculation method, the calculation process is allocated to the cluster machines and each machine handles different parts of the calculation process, eventually to merge the results for each part. The main work is described as follows:(1) In the map-matching system, this paper, propose a new way which called HashMap gridding index algorithm. This algorithm makes the time complexity is reduced to O(1) and solve the problem of the sharp decline in the efficiency of query of traditional four fork tree indexing algorithm which is uniformly distributed in the space object. By dividing two meshes and a central region, the matching accuracy has been improved.(2) Importing elevation height information of the map-matching system, the map-matching algorithm, split into elevated/non elevated matching, by determining where the gridding buffer contains an elevated section of information when matching points is passing, And then selecting the matching algorithm, This program improved the shortcomings of the traditional method in the treatment of overlapping between viaduct and ground roads, so as to further improve the matching accuracy degree.(3) To the problem of time-consuming in massive floating car data of the traditional desktop computing model for map matching, this paper, based on the Hadoop cloud platform, through the Map/Reduce programming model, computing the large-scale floating car data which distributed in parallel, quickly and efficiently realizing the map matching.(4) Based on single track matching test data, computing overlapping of viaduct and ground path matching and testing the large-scale floating car data, this paper got that the data shows better in the algorithm in accuracy and computational efficiency. Therefore, This paper has not only theoretical innovation, but also high practical application value.