Design And Implementation Of Spatio-temporal Data Storage And Index Structure

Spatio-temporal DBMS (database management system) is a DBMS which can handle the temporal data and the spatial data. How to combine the temporal attributes with the spatial attributes is a critical problem to implement the spatio-temporal DBMS. With the development of GPS,GIS, Spatio-temporal DBMS becomes one of the most important branches of database technology.The SAILII system is a prototype of Spatio-temporal DBMS by extending the postgreSQL, and it can support Bitemporal&2Dimension Spatial data model.The management of data storage of the SAILII System can support SAILII data model very well, which includes supporting different Spatio-Temporal datatypes, supporting size-fixed and size-variable records, and organizing the physical blocks effectively. SAILII stores the size-variable attributes in the other blocks, which makes the records size-fixed. SAILII stores the offset table from the beginning of the block, and stores the records from the ending of the block, which makes the slide of the records in the block easy. Auto mix writing pointers used in SAILII solves the problem of moving blocks between the main memory and the second level storage devices.The TDLRTree (Temporal Double Level RTree), the index structure of the SAILII System, makes the query for Bitemporal&2Dimension Spatial data more efficient. It indexes the bitemporal data by HRTree, and then indexes the still and the moving spatial objects of the spatial state of one moment by Double Level RTree. In TDLRtree, still object is the container of moving object, and for each object, there is a bi-directional chain, which link the past and the future spatial states of the object. The TDLRtree can support more query types than traditional spatio-temporal indexes, and in some aspects, its query efficiency is better than traditional ones.