Research And Application Of Distributed Spatio-temporal Data Index Mechanism

With the popularization of smart devices and the development of cloud computing and mobile Internet, the navigation positioning and social networks which are based on the LBS(Location Based Service) are widely used. The LBS systems should provide the information timely for users according to the frequently updated location data. These systems usually choose moving object database(MOD) for data storage and management, which provide the query for the past, current and future locations of moving objects. On one hand, although the traditional MOD based on relation model could provide more mature location data storage, indexing and query methods, its scalability and real-time performance could not adapt to the management and the query of a large mount of location data. On the other hand, despite the cloud database systems based on Key-Value mechanism could perform better in scalability than traditional MOD based on relation model, it could not provide efficient access method to support querying the locations of moving objects in past, current and future.To solve these problems described above, a parallel indexing the past, current and future locations of moving objects(PIPCF) is proposed in this paper. PIPCF is a hybird structure which consists of two parts, the indexing and cache parts. The indexing part firstly uses the Quad-Tree to split the spatial area to be the static spatial index, and then uses the R-Tree to index the location data in each sub area of the static spatial index. PIPCF also introduces a hash table based on the memory to manage the prediction trajectories of the moving objects, so as to improve the update efficiency. While, the cache part is mainly used to cache index structure data to reduce the impact of disk and network I/O on index performance when the index is frequently updated. The creating and maintaining algorithm and the query algorithm which is for querying the past, current and future locations of the moving objects based on the PIPCF are also proposed in this paper.This paper finally analyzes the update efficiency of the PIPCF and the location querying performance through a real traffic data set. The experimental results show that PIPCF could perform well in parallel mode and improve the querying performance of the locations of moving objects in past, current and future.