Multidimensional Index Research Based On Two Division Methods

With the rapid development of medicine, molecular biology, network, Gis spatiotemporal databases have been the focus of considerable research activities over a significant period. Multidimensional data index is a way to index the eigenvector in the multidimension space. According to the data's division and organization, multidimensional data indexes can be divided to two classes: indexes based on space, indexes base on data .The multidimensional data index has been a very active research area over the last few years.Quadtree is the typical one of the indexes based on space who is characterized with easy building, quick queries, easy operation and etc. But the scope of the object must be known before the index is built, and the object conditions affect the efficiency of storage and query very much. Theory and practice prove that encoding the leaf nodes of quadtree reduces the quantity of the nodes and enhances the storage efficiency. A new way of Leaf-coding is designed, based on the coding, the algorithm for neighbor searching of leaf-coding quadtree is implemented. The method enhances the storage efficiency,and it can also make the neighbor searching using bitwise operating available at the leaf node level, enhance the query efficiency.Hilbert curve is the typical one of the indexes based on data which has good properties of space cluster and self-similarity. But tradional Hilbert curve can only be operated in 2-dimension or 3-dimension,which limits the applications of the Hilbert curve as a high-dimension index.Researching the 2-dimention and 3-dimension Hilbert curve's properties,a new way of generating the cell of Hilbert curve is proposed ,which breaks the limit of the dimension and makes reseach of high-dimension Hilbert curve possible.Based on the research above,a experiment flat is designed which is used to test the indexes . It is an opened structure, and users can add their indexes into it by some rules anytime.