The Research Of Temporal Index Technique And Algorithm

Most data managed by existing database application systems is time referenced. Data warehouse is a good example. Often, the data with one of the following two temporal aspects is called temporal data. Two temporal aspects of data are valid time and transaction time. And the data with the above two temporal aspects is called bitemporal data. Like spatial data, bitemporal data has associated two-dimensional data regions. The most important data is the current-time-relative data such as Now and UC in temporal database. But over the last several decades development in the field of the temporal database, only little work has addressed the indexing of current-time-relative data while most research has focused on indexing general temporal data, which is perhaps one of the reasons why commercial RDBMS does not support the temporal data index technique. Thought the 4R-tree index yields a performance that is competitive with the best existing GR-tree index in theory, it may be implemented as a layer on top of any DBMS that supports R-trees, not to require an extension of the DBMS's kernel. However, To develop a more robust and more efficient access method for temporal database than the existing ones is an important research question in the temporal database's field.This paper proposes a new index technique, known as G2R-tree index, that can process efficiently all types of bitemporal data. Moreover its performance is competitive with the 4R-tree index and the GR-tree index. The G2R-tree index combines the 2R-tree index, the GR-tree index, the G4R-tree index, and expands the query function of the 2R-tree index technology on the basis of the 2R-tree index, and supports the query of the future case. The G2R-tree index has the advantages of the G4R-tree Index and the 2R-tree index. Simultaneously it overcomes the flaws of the 2R-tree index and the G4R-tree index so it only need to access two R-trees at the query time which is different to the 4R-tree index and the G4R-tree index that need to access four R-trees, thus it reduces the I/O operations greatly and it better resolve the inevitable problem of the 4R-tree Index that I/O operation is increased tempestuously at the query time. This paper improves the indexing algorithm and the access algorithm in order to enhance the performance of the G2R-tree index with processing all types of bitemporal data.This paper introduces the bitemporal data transformation of the G2R-tree index and the query transformation on the two R-trees of the G2R-tree index in detail. It presents the operation algorithms of the G2R-tree index and indicates the correctness of the transformed G2R-tree index on query through some theorem. Finally this paper proves the high-effectiveness and feasibility of the G2R-tree index by the contrasted performance experiments of the G2R-tree index, the 2R-tree index, the GR-tree index and the 4R-tree index.