摘要
空间索引是空间数据库的关键组成部分,其性能的优劣直接决定着空间数据操作的效率。为此,在分析了现有各种空间索引的基础上,将分布并行处理技术与空间索引相融合,提出了一种 DPsIR^+树。DPsIR^+树借助繁衍和返祖,动态分割空间槽,并将它们映射到多个节点机上。每个节点机再将其对应空间槽中的空间对象组织成 R 树,并将R树分裂成多个残枝,将残枝并行存入本地 MultiDisk 中;在内存中则按 R-link 组织空间对象,按 R^+处理节点溢出。实验结果表明 DPsIR^+树具有良好的查询特性。
Spatial index is important part of spatial database, whose performance is vital to efficiency of spatial opera tion. After various spatial indexes were analyzed and distributed and parallel processing was introduced into spatial index, this paper has come up with a DPsIR^+ tree, which splits space into multi-slots by multiplying and reverting and maps these slots to sites in distributed and parallel system. Each site constructs the spatial objects in their spatial slot into an R tree, splits the R into stumps and stores the stumps into multidisk. In main memory, site builds Rlink tree and deals with overflow as R^+ tree does. Experiment results show that DPsIR^+ tree behaves well during spatial query.
出处
《计算机科学》
CSCD
北大核心
2006年第2期121-126,共6页
Computer Science
基金
四川省科技攻关项目(02GG006-018)基金资助。