A generic algorithmic framework for aggregation of spatio-temporal data

Spatio-temporal databases are often associated with analyses that summarize stored data over spatial, temporal or spatio-temporal dimensions. For example, a study of traffic patterns might explore average traffic densities on a road network at different times, over different areas in space, and over different areas in space at different times. The importance of temporal, spatial and spatio-temporal aggregation has been reflected in a significant number of proposals for algorithms for efficient computation of specific kinds of aggregation. However, although such proposals may be effective in particular cases, as yet there is no generic framework that provides efficient support for the wide range of partitioning and aggregation operations that a spatio-temporal database management system might be expected to support over both stored and derived data. This paper proposes an algorithmic framework that can be applied to many different forms of aggregation, and presents the results of performance studies on an implementation of the framework. These show that the framework provides a scalable solution for the many cases in which the aggregations required over stored and derived data may be widely variable and unpredictable.