Please use this identifier to cite or link to this item:
Title: A spatial clustering based method for partitioning of vector spatial data for parallel computation
Other Title: 基于空间聚类的矢量空间数据并行计算划分方法
Authors: Qiu, Q
Fang, L
Yao, X
Fang, J
Issue Date: 2015
Source: 高技术通讯 (High technology letters), 2015, v. 25, no. 4, p. 327-333
Abstract: 为了解决并行矢量空间分析在数据划分阶段的负载均衡问题,研究了矢量空间数据的划分,提出了一种基于空间聚类思想的矢量空间数据划分方法。该方法充分考虑矢量空间数据规模以及空间邻近性特征对并行空间分析算法效率的影响,首先采用空间填充曲线对二维空间数据进行编码,保证空间要素邻近性特征;然后用空间要素集合对空间要素流进行填充,从而确保各个子任务集中的要素数据规模相对均衡。以并行叠加分析中点面、线面、面面叠加操作为例,设计了对比实验。实验结果表明,该方法能够有效提高以线、面要素为操作对象的并行算法负载均衡度和提高并行算法整体运行效率。
The partitioning of vector spatial data was studied, and a new data partitioning method based on spatial clustering was proposed to deal with the load balancing problem in the data partitioning stage of parallel vector spatial analysis. This method fully considers the influence of the vector spatial data size and spatial proximity on the efficiency of the algorithm for parallel vector spatial analysis. Firstly, it uses space filling curves to encode the two-dimensional spatial data to keep the characteristic of spatial proximity. Secondly, it fills the features to the spatial feature box to ensure the balance of the feature sizes in each slaver processing. The operations of point-to-surface, curve-to-surface and surface-to-surface overlay were used as the examples to design the contrast test. The experimental result proved that this proposed method improved the load balancing degree and the whole efficiency of the parallel algorithm on the curve and surface spatial data.
Keywords: Feature box
Hilbert curve
Load balancing
Parallel computing
Vector data
Publisher: 中国学术期刊(光盘版)电子杂志社
Journal: 高技术通讯 (High technology letters) 
ISSN: 1002-0470
DOI: 10.3772/j.issn.1002-0470.2015.04.001
Rights: © 2015 中国学术期刊电子杂志出版社。本内容的使用仅限于教育、科研之目的。
© 2015 China Academic Journal Electronic Publishing House. It is to be used strictly for educational and research use.
Appears in Collections:Journal/Magazine Article

View full-text via PolyU eLinks SFX Query
Show full item record

Page view(s)

Last Week
Last month
Citations as of Nov 24, 2020

Google ScholarTM



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.