Massive Spatio-Temporal Mobility Data: An Empirical Experience on Data Management Techniques

The technological improvements within the Intelligent Transportation Systems, based on advanced Information and Communication Technologies (like Smartphones, GPS handhelds, etc.), has led to a significant increase in the availability of datasets representing mobility phenomena, with high spatial and temporal resolution. Especially in the urban scenario, these datasets can enable the development of "Smart Cities". Nevertheless, these massive amounts of data may result challenging to handle, putting in crisis traditional Spatial Database Management Systems. In this paper we report on some experiments we performed to handle a massive dataset of about seven years of parking availability data, collected from the municipality of Melbourne (AU), being about 40 GB. In particular, we describe the results of an empirical comparison of the retrieval performances offered by three different off-the-shelf settings to manage these data, namely a combination of PostgreSQL + PostGIS with standard indexing, a clustered setup of PostgreSQL + PostGIS, and a combination of PostgreSQL + PostGIS + Timescale, a storage extension specialized in handling temporal data. Results show that the standard indexing is by far outperformed by the two other solutions, which anyhow have different trade-offs. Thanks to this experience, other researchers facing the problems of handing these kinds of massive mobility dataset might be facilitated in their task.