Towards a Trajectory Data Warehouse

The widespread diffusion of modern technologies such as low-cost sensors, wireless, ubiquitous and location-aware mo-bile devices, allows one to collect an overwhelming amount of data about trajectories of moving objects. Such data are usually produced at different rates, and arrive in streams in an unpredictable and unbounded way. In this paper we discuss how standard data warehousing tools can be used to store trajectories and to compute OLAP operations over them. We construct a data cube where the dimensions are the spatial coordinates and the time, dis-cretized according to a regular grid. The measures of inter-est are the number of objects and the distance travelled by such objects for any (spatio-temporal) cell of the data cube. According to the way trajectories are represented in the data warehouse the measures can be possibly affected by errors. We discuss how such errors can be influenced by the diverse spatio-temporal granularities vs. the speeds and sampling rates of the moving objects, and we study the accuracy of the computed measures vs. the amount of pre-processing of raw data. Finally, we analyze whether the usual OLAP ag-gregation operations, like sum and average, may introduce further errors when rolling-up