Computational resources. IoT devices have limited processing
resources that do not allow on-site data processing. Data
collected is usually transmitted to more powerful nodes where
aggregation and processing is possible, but scalability is challenging
to achieve without a proper infrastructure. The unlimited
processing capabilities of Cloud and its on-demand model
allow IoT processing needs to be properly satisfied and enable
analyses of unprecedented complexity [27], [47]. Data-driven
decision making and prediction algorithms would be possible
at low cost and would provide increasing revenues and reduced
risks [56]. Other perspectives would be to perform real-time
processing (on-the-fly) [50], [27], to implement scalable, realtime,
collaborative, sensor-centric applications [32], to manage
complex events [50], and to implement task offloading for
energy saving [53].