CAMS: OLAPing Multidimensional Data Streams Efficiently

Abstract

In the context of data stream research, taming the multidimensionality of real-life data streams in order to efficiently support OLAP analysis/mining tasks is a critical challenge. Inspired by this fundamental motivation, in this paper we introduce CAMS ( C ube-based A cquisition model for M ultidimensional S treams ), a model for efficiently OLAPing multidimensional data streams . CAMS combines a set of data stream processing methodologies, namely (i ) the OLAP dimension flattening process , which allows us to obtain dimensionality reduction of multidimensional data streams, and (ii ) the OLAP stream aggregation scheme , which aggregates data stream readings according to an OLAP-hierarchy-based membership approach. We complete our analytical contribution by means of experimental assessment and analysis of both the efficiency and the scalability of OLAPing capabilities of CAMS on synthetic multidimensional data streams. Both analytical and experimental results clearly connote CAMS as an enabling component for next-generation Data Stream Management Systems .