【作者】 周芹； 钟耳顺； 黄耀欢； 郭会；

【Author】 Zhou Qin1, 2), Zhong Ershun1), Huang Yaohuan3) , Guo Hui1, 2) (1 Institute of Geographic Sciences and Natural Resources Research Chinese Academy of Sciences China, 100101) (2 Institute of Geographic Sciences and Natural Resources Research, CAS, Beijing 100039) (3 Department of Water Resources, China Institute of Water Resources and Hydropower Research, Beijing 100044, China)

【机构】 中国科学院地理科学与资源研究所； 中国水利水电科学研究院水资源研究所；

【摘要】 在基于纯关系数据库的空间数据库中,空间索引通常由空间数据引擎(Spatial Data Engine)维护,R树索引由于其优良的性能被广泛应用于空间数据引擎中。在多用户并发环境下,R树的并发控制,算法复杂,如改变R树存储结构的R-link树[1]及其改进算法[3]、纯粹基于锁的控制算法[2]等,不适合海量空间数据的并发操作。基于以上原因本文提出适用于客户端模式空间数据库引擎并发控制的空间索引结构——CQR树(Concurrent Quad-tree&R-tree),将静态R树与四叉树相结合,采用四叉树编码与空间对象绑定的方式管理被编辑过的对象,仅在删除叶子结点包中的对象时对相关索引包加锁,缩短系统响应时间;算法简单易于实现,在保证空间查询效率的前提下解决多客户端并发操作的问题,同时降低索引的维护难度。更多还原

【Abstract】 R-tree is incapable of managing spatial objects in concurrent environment. This article proposes the CQR-tree (Concurrent Quad-tree & R-tree) to satisfy this situation, which is easy to integrate with existing R-tree systems. The article is organized as following: firstly, we point out the fault of the R-tree in concurrent environment and the limitation of the R-link tree in the special system; secondly, we introduce the CQR-tree to solve the problem; thirdly, we list the algorithms of the CQR-tree, including insert, add and delete algorithms, and the query strategy; then, some experimental results confirm that proposed CQR-tree performs well in concurrent environment; finally, the results and some future work are pointed out.更多还原