An Efficient Distributed Index for Geospatial Databases

Abstract

The recent and rapid growth of GPS-enabled devices has resulted in an explosion of spatial data. There are three main challenges for managing and querying such data: the massive volume of data, the need for a high insertion throughput and enabling real-time spatial queries. Although key–value store databases handle large-scale data effectively, they are not equipped with effective functions for supporting spatial data. To solve this problem, we propose an efficient spatial index structure based on HBase, a standard key–value store database. We first use Geohash as the rowkey in HBase to sustain high insert throughput. We present a novel data structure, the binary Geohash rectangle-partition tree, that partitions data into subrectangles, then add these subrectangles into an R-Tree to support spatial queries. Our experiments demonstrate high scalability and an improved performance with spatial queries, when compared with a state-of-the-art system. They also show a good real-time query-processing capability, with response times of less than one second.