Indexing Query Regions for Streaming Geospatial Data

This paper introduces the Dynamic Cascade Tree (DCT), a structure designed to index query regions on multi-dimensional data streams. The DCT is designed for a stream management system with a particular focus on Remotely-Sensed Imagery (RSI) data streams. For these streams, an important query operation is to efficiently restrict incoming geospatial data to specified regions of interest. As nearly every query to an RSI stream has a spatial restriction, it makes sense to optimize specifically for this operation. In addition, spatial data is highly ordered in it’s arrival. The DCT takes advantage of this trendiness. The problem generalizes to solving many stabbing point queries. While the worst case performance is quite bad, the DCT performs very well when the stabbing point exhibits certain trending characteristics that are common in RSI data streams. This paper describes the DCT, discusses performance issues, and provides extensions of the DCT.

• Type: Conference
• Year: 2004
• Where: STDBM
• Authors: Quinn Hart, Michael Gertz
• Keywords: Data Streams, Database, DCT Takes Advantage, Multi-dimensional Data Streams, STDBM 2004