吉林大学学报(理学版)

• 计算机科学 • 上一篇    下一篇

并行环境下基于图着色理论的空间数据部署

殷君茹1, 唐小明1, 李惺颖2, 卜祥亮3   

  1. 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
  • 收稿日期:2014-10-28 出版日期:2015-05-26 发布日期:2015-05-21
  • 通讯作者: 唐小明 E-mail:tangxm@caf.ac.cn

Graph Coloring Based Spatial Data Placement towards Parallel Computing System

YIN Junru1, TANG Xiaoming1, LI Xingying2, BU Xiangliang3   

  1. 1. Research Institute of Resource Information Techniques, Chinese Academy of Forestry, Beijing 100091, China;2. RS, GIS, GPS Technology Research & Development Center, Guangxi Forest Inventory & Planning Institute, Nanning 530011, China; 3. College of Soil and Water Conservation, Beijing Forestry University, Beijing 100083, China
  • Received:2014-10-28 Online:2015-05-26 Published:2015-05-21
  • Contact: TANG Xiaoming E-mail:tangxm@caf.ac.cn

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.

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

Abstract:

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.

Key words: spatial data placement, data granularity, parallel computing system, graph coloring theory, load balancing

中图分类号: 

  • TP391