A Research Of Spatio-Temporal Object Query Processing Technology Oriented To Column Storage Model

With the development of global positioning technology and the progress of information collection technology, spatio-temporal data grow rapidly. In the face of massive, diverse, complex spatio-temporal data, it becomes a hot issue between domestic and foreign scholars that how to store and manage the spatio-temporal data efficiently. The traditional centralized spatio-temporal index within single server environment is difficult to meet the cost of the massive spatio-temporal data storage and query. Therefore building a distributed spatio-temporal index has become an inevitable trend. Compared with traditional row-oriented relational database, the Column-oriented database has greater advantages in dealing with massive data. Its parallel data access model measured in columns and properties such as high availability, parallel mode, easy expansion, dynamic load balancing etc, provide a solution for the management of massive spatio-temporal data.This paper designed two kinds of spatio-temporal index based on the basic theory of database of column-oriented storage model and spatio-temporal index, and described the building of two kinds of index, the design of query algorithm, the maintenance and optimization of index. Finally, used HBase column-oriented database and made an evaluation of the two kinds of index performance through the experiment and analysis. The main contributions of this paper are as follows:(1) This paper proposed a structure of spatio-temporal index oriented to column storage model. Based on the deep research of HBase column-oriented database and spatio-temporal index, combined with the distributed environment characteristics of column-oriented database and the previous spatio-temporal indexing technologies, proposed a structure of spatio-temporal index oriented to column storage model, selected and improved the technologies related to the construction of spatio-temporal index to adapt to the new distributed and column-oriented environment. This laid a good foundation for the construction of the spatio-temporal index.(2) This paper proposed a "no table" structure of spatio-temporal index oriented to column storage model. Throuth the reserach of the spatio-temporal data storage model in column-oriented database, according to the spatio-temporal data model, proposed three design mode of rowkey and designed a HBase spatio-temporal data storage structure which corresponded with the rowkey. Then this paper built a "no table" structure of spatio-temporal index oriented to column storage model by comparing and selecting the optimal spatial transformation combination mode of rowkey. Described the process of building the index, designed the range query and the nearest neighbor query algorithm of the spatio-temporal index, and offered some maintenance and optimization strategy.(3) This paper proposed a "two level table" structure of spatio-temporal index oriented to column storage model. By using the optimal rowkey and the corresponding data storage structure, this paper built a "two level table" structure of spatio-temporal index oriented to column storage model with time level index table and time space index table. Describe the building of two level index table and the indexing process, designed the range query and the nearest neighbor query algorithm, and offered some maintenance and optimization strategy.(4) This paper made a series of experiments for the two kinds of index and presented an evaluation of the index performance. Used HBase column-oriented database and tested the performance of two kinds of spatio-temporal index based on HBase by a series of experiments, Analyzed the the experimental results and offered objective evaluation.