An Architecture to Process Massive Vehicular Traffic Data

Fostered by the "big data" hype in mobility, many research efforts have been aimed at improving techniques to model vehicular traffic patterns for mobility prediction. Nevertheless, from a practical stance, the industry still faces many technological challenges in bringing solutions on the market. Especially the scalability and performance of such systems raise major concerns, given the amount of spatio-temporal data to be processed. The common approach in dealing with these issues is to introduce constraints and/or simplifications on both the spatial component of the data and on the employed algorithms, leading to results that are somehow limited. To overcome these issues, in this paper we report on our experiences and our approaches in providing a solution that meets industrial needs with the aim to leverage the computational and storage capabilities of the Cloud to handle massive dataset for providing vehicular traffic predictions. In particular, we present an approach to deal with real-world datasets to facilitate the knowledge discovery process from this data while matching the business constraints given by the industrial use case.