Large-scale Trajectory Management And Clustering Analysis Method Based On NoSQL

In recent years, as the widespread use of civilian GPS(Global Positioning System) in mobile terminals, and the development and popularization of Location-Based Services and Mobile Social Networks, lots of trajectory are accumulating in our daily life and serving for different types of applications,management, indexing and analysis of large-scale of trajectory become very important.In this paper, the research object is trajectory of moving object, and its main purpose is storage, query, and analysis for moving objects.This paper solves several key problems of moving object database from three aspects.(1) Trajectory Management. This paper analyses moving object model, uses non-relationship database NoSQL to store large-scale trajectory of moving object, and does performance optimization on insertion and query of large-scale trajectory;(2)Trajectory Index. This paper proposes trajectory mixing index to improve trajectory query efficiency. It designs a mixing index with R-tree, hash table and B tree,achieves the interactively internal-external storage mechanisms and the index cache mechanisms, and describes hierarchical query algorithm of trajectory index in detail.Based on trajectory data sets from spatio-temporal generator, the trajectory mixing index of this paper does experimental comparison with classic TB tree in index created performance and query processing performance. This paper analyses the experimental data, and experiment shows that the trajectory mixing index has well index created performance and query processing performance;(3) Trajectory Clustering Analysis. This paper does problem definition in trajectory model, trajectory partition and trajectory similarity based on data mining. Based on spatio-temporal generator to provide network data of Olburgen city road and simulative trajectory data,and trajectory R-tree mixing index to reduce search space in clustering analysis, this paper proposes large-scale trajectory cluster method and analyses the clustering results through experimental tests. Experimental analysis shows that large-scale trajectory clustering analysis method has good clustering capability and efficient clustering analysis.