A Method of Spatial Data Partition for Efficient Parallel Computing of Topological Relations
Yang Yi-zhou · Geography and Geo-Information Science · 2013
Referring to large-scale spatial data,the computational efficiency is low for serial algorithm of topological relations.In order to improve the efficiency,parallel algorithms are introduced,which employ more computing resources with the parallel computing environment.Most essential and basic part of the algorithms is data partition.After analyzing the characteristics of parallel algorithm of topological relations,this paper proposes a method of spatial data partition for load balance and task balance in parallel computing environment.The experiment result shows that the speed ratio increases linearly with the number of process,and that the parallel efficiency of the algorithm stabilizes around 80%not decreases with the number of process.An efficient strategy of spatial data partition is introduced for high efficiency parallel computing of topological relations in this paper,which can be employed in parallel software development and data service for large-scale spatial data on parallel commutating environment.