Rectilinear Range Query Processing on SpatialHadoop Platform

Querying information from big geographic data attracts a lot of attentions. As the data are extremely large, it still takes a lot of computing time even by parallel computing with Hadoop/MapReduce and cannot meet the requirement of prompt response. For storing spatial big data, putting geographically close data together are easier for indexing as well as information querying, which is most adopted in the nowadays database systems like SpatialHadoop or Hadoop-HIS. There are a heavy demand on querying data within a polygonal administrative re-gions like country and city. However, most database management systems do not support polygon query. In the paper, we study the problem of polygon query in the big geographic data and adopt a filtering-refinement strategy to reduce the computing time. Specifically, since it is easy to determine whether a data point is within a rectilinear polygon, we propose to use a rectilinear polygon to cover the inner area of a polygon region so as to filter out most data at the first stage. Then, we use the traditional ray-casting algorithm to examine the rest data within the rectangles that are crossed by the polygon border. To verify our design, we conduct experiments by using the earthquake datasets and the results show that our approach significantly reduce the computing on the Hadoop or Spatial Hadoop platforms, especially when the polygon is complex.