Abstract: A novel method for index-building of TPR*-tree is proposed, which splits the moving object set at root node level by using velocity vector. According to the value of velocity vector, the moving objects are clustered into different sub-nodes, which then constructs the TPR*-tree. The records of moving objects are stored by using overflow bucket at root node level, and the index of TPR*-tree is inserted and updated with a batch, which decreases the cost of insertion maintenance. Experimental results show this method is feasible.

Key words: TPR*-tree, construction algorithm, overflow bucket, insertion delay update

摘要： 提出一种新的TPR*树索引构建方法，在根节点层利用速度矢量对移动对象集进行划分，根据速度矢量的大小将移动对象聚集到不同子节点中，并逐层构建TPR*树。在根节点层用溢出桶存储插入的移动对象记录，同时对TPR*树索引进行批量插入更新，以减少其插入更新维护的代价。实验结果表明，该方法是可行的。

关键词: TPR*树, 构建算法, 溢出桶, 插入延迟更新

CLC Number: 

• TP311.12