并行环境下基于图着色理论的空间数据部署  

Graph Coloring Based Spatial Data Placement towards Parallel Computing System

作　　者：殷君茹[1] 唐小明[1] 李惺颖 卜祥亮 

机构地区：[1]中国林业科学研究院资源信息研究所,北京100091 [2]广西林业勘测设计院3S技术研究与开发中心,南宁530011 [3]北京林业大学水土保持学院,北京100083

出　　处：《吉林大学学报（理学版）》2015年第3期525-530,共6页Journal of Jilin University:Science Edition

基　　金：国家高技术研究发展计划863项目基金(批准号:2012AA102001);国家林业公益性行业科研专项基金(批准号:201304215)

摘　　要：在面向计算部署到数据节点端执行的分布式并行环境下,提出一种基于图着色理论的适用于矢量空间数据的部署方法,将空间数据粒度的部署问题转化为图顶点着色的过程,提高了任意空间区域的信息查询效率.给出基于图着色理论的数据部署方法,并通过节点的任务量进一步改进算法,使得该算法可实现海量空间数据粒度的离散化部署,提高了空间数据检索和查询的并行化程度,充分利用了并行计算资源.An algorithm suitable for spatial vector data placement based on graph coloring theory was presented in the parallel system of computing distributed to data nodes. The deployment problem was transferred into graph vertex coloring problem, and the information query efficiency of any spatial area was thus improved. Moreover, the algorithm based on graph vertex coloring problem was proposed and improved by the task of nodes. This algorithm can achieve discrete deployment of massive spatial data granularity and storage load balance of the nodes, improve the degree of parallelism spatial data retrieval and query, and make full use of parallel computing resources.

关 键 词：空间数据部署 数据粒度 并行环境 图着色理论 负载均衡 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]