基于PQR-tree的空间查询代价模型

空间信息处理和地理信息系统等领域的数据管理涉及到海量、高维空间数据对象的处理。本文针对传统数据索引结构在处理这类空间数据时所存在的内存使用过大、I/O消耗过多等问题,通过改进选择查询的代价模型,给出了基于PQR-tree的查询和代价模型,以提高空间数据查询的性能。提出了基于PQR-tree的三阶段并行查询的方法,分别在任务创建、分配、执行阶段进行优化。提出在任务创建和任务分配阶段应用于空间查询中过滤和精炼阶段的有效算法。测试表明,本文算法在处理各种不同分布类型数据集过程中有效降低了空间数据处理对时间和空间的代价和需求,并且并行机制下的代价模型在预测和评估方面也具有较好的精确度。

A Cost Model for Spatial Queries Based on PQR-tree

The operation of spatial data objects is frequently involved in many applications with mass and high dimensional data sets,such as spatial information and GIS.The traditional index has disadvantages with too much memory and I/O consumption in handling this kind of queries.This paper presents the PQR-tree and cost model for spatial queries using PQR-tree.With the improvement of the select query cost model,the corresponding query and cost model are given based on PQR-tree.In order to improve the performance of the query of spatial data,a three-phase parallel query method is presented.And the optimization is executed respectively in the task creation,the distribution,and the execution stages.The effective filtering and refining algorithm can improve the performance in task creation and task distribution which are applied to spatial query.The experiment shows that the model has a degree of accuracy,and the above methods can reduce the cost of time and spatial requirements with different data sets.Given that parallel mechanism treats targets efficiently,this paper proposes a parallel mechanism cost model by using parallel query methods.