A ONCE-Updating Approach on Moving Objects

Abstract

Indexing methods that manage current states of moving objects in a database are confronted with keeping their indexes up to date. Numerous objects that move continuously can invoke many their updates at a time constantly. R-tree based indexing methods promise to be efficient and easy to be implemented and deployed to real applications. Nevertheless, recent updating improvements for R-tree have been not either designed or appropriate to a moving object environment. Therefore, in this paper, we propose a ONCE-Updating approach applied for indexing methods developed from the R-tree. The idea is that instead of incurring the high cost of updating an indexed data on many new states at a timestamp one by one, an indexing method can easily update the data on all the states just once. In our experiments, in addition to much improving update performance, the approach even increases quality of queries extracted from indexed data. This approach can be used to support methods managing data from traffic surveillance systems.