Data science applications to connected vehicles: key barriers to overcome

As our environments become more connected in general, Intelligent Transportation Systems will play a central role in our cities and across borders, forming part of a new vision of “mobility as a service.” Connected vehicle technology, with a prominent role in Intelligent Transportation Systems, will be capable of generating huge amounts of pervasive and real time data, at very high frequencies. These streaming data are the common type of data produced by Connected vehicles, and their analysis is of paramount importance for applications improving road safety, effective service delivery, eco-driving, traffic regulation and pollution reduction. This study aims to characterize this type of data from an analytical perspective, as well as to pose the challenges Data science faces in extracting knowledge from them in real time. Data generated by sensors and actuators in Connected vehicles include noisy, anomalous, redundant, rapidly changing, correlated and heterogeneous data. In such a context, numerous techniques have been proposed to adapt the data analytics in batch learning to these new dynamic and evolving streaming data, which are produced in huge volumes and transmitted at high velocity. The Internet of Vehicles has the potential to provide a pervasive network of Connected vehicles, smart sensors and road infrastructures, and big data has the potential to process and store that amount of data and information. Modelling, predicting, and extracting meaningful information in reasonable and efficient ways from big data represent a challenge for Data science in Connected vehicles.

Media Info

  • Pagination: 34p

Subject/Index Terms

Filing Info

  • Accession Number: 01660768
  • Record Type: Publication
  • Source Agency: ARRB
  • ISBN: 9789279770418
  • Files: ATRI
  • Created Date: Feb 20 2018 10:43AM