Efficient Distributed Spatial Semijoins and Their Application in Multiple-Site Queries

Applications exist today that require the management of distributed spatial data. Since spatial data is more complex than non-spatial data, performing distributed queries on it requires the consideration of both local processing (i.e. CPU and I/O) time and data transmission cost. To reduce these costs, one can use a distributed spatial semi join as it eliminates unnecessary objects before their transmission to other sites and the query site. In this paper, we propose both new approaches for representing the spatial semi join in a distributed setting, and their use for processing distributed queries consisting of any number of sites. We have tested our algorithms for four sites, which are a part of an actual working distributed system. We compare our algorithms with respect to data transmission cost, CPU time, I/O time and false positive results. We show that our algorithms are superior in many cases at optimizing the above criteria.