In this paper, we address the maintenance of spatial semijoin queries over continuously moving points, where points are modeled as linear functions of time. This is analogous to the maintenance of a materialized view except, as time advances, the query result may change independently of updates. As in a materialized view, we assume there is no prior knowledge of updates before they occur. We present a new approach, continuous fuzzy sets (CFS), to maintain continuous spatial semijoins efficiently. CFS is compared experimentally to a simple scaling of previous work. The result is significantly better performance of CFS compared to previous work by up to an order of magnitude in some cases.
Glenn S. Iwerks, Hanan Samet, Kenneth P. Smith