Perfect KDB-Tree: A Compact KDB-Tree Structure for Indexing Multidimensional Data

Many literatures have improved the techniques of traditional KDB-tree and its variants for acquiring some performance enhancement. However, they all suffer from the low storage utilization problem caused by their imperfect "splitting policies." Frequent splits do not only increase the size of index structure but also deteriorate the performance of the system. A new insertion algorithm with the new splitting policy was proposed, which can insert data entries in the leaves as much as possible to increase storage utilization up to nearly 100%. Analytical and experimental results show that our indexing method outperforms the traditional KDB-tree and its variants