The Map Matching Algorithm Based On Hadoop

With the rapid development of urban road traffic related to the rapid growth of the number of vehicles and the amount of traffic data of the growth of the geometric series, related to car data processing accuracy, reliability and real-time performance are put forward higher requirements. The current study of vehicle data processing algorithm is more short intervals of small sample data, on the basis of the relative lack of longer time interval data matching algorithm. At the same time, the existing database management technology has been difficult to meet the growing mass of floating car data management and application requirements. To solve these problems, this paper proposes a realization of large sample based on Hadoop floating car GPS data processing of map matching algorithm. Paper main research work is as follows:1) based on the existing non-relational database and traditional relational database of data organization and the comparison of application scope, non-relational database for huge amounts of data management are pointed out the advantages and disadvantages, studies the Hadoop of the specific function of each component, the basic principle, mainly study the graphs programming thought and distributed processing, lay a theoretical foundation for the following study.2) according to the vehicle and road geometry relationship, for all kinds of typical reviewed road map matching algorithm, this paper introduces the existing large sample vehicle data map matching algorithm is implemented, points out the necessity of applying cloud computing to road map matching, and the problems existing in the current study.3) use and improve the existing road compression method to realize the road data preprocessing, a detailed analysis of the spatial factors(including Angle, distance, connectivity, time factors such as upload interval, speed, etc.) and history matching results on the result of matching, improved the calculation of time factor, are given a road map matching algorithm based on matching degree feedback, and is suitable for parallel computing of the matching algorithm of Hadoop optimization.4) using virtual machine set up Hadoop platform, for the proposed based on matching degree feedback path matching algorithm, the simulation experiment respectively from matching the effect, the algorithm performance, on the application of the platform degrees are verified. Through experiment contrast before and after the match the algorithm of vehicle location changes prove that the matching algorithm was adopted to realize a long interval location data matching the feasibility and effectiveness of road.