Ricerc@Sapienza

This proposal plans the activity for the second year of the (two-year) "Vehicular Fog for energy-efficient QoS mining and dissemination of multimedia Big Data streams" (V-Fog) project. The latter aims at defining, designing and validating integrated resource-management and data-mining distributed self-adaptive algorithms for vehicular Fog networks. The final goal is the energy-efficient support of real-time Big Data streaming Future Internet Applications, such as multimedia human activity recognition and infotainment interactive services (e.g., VTube). For this purpose, the actual (still unexplored) transport capabilities promised by the novel paradigm of multipath-TCP will be investigated, while a cognitive approach will be pursued for the integrated design of the V-Fog architecture. The project is motivated by the consideration that the delivering cloud-supported real-time Big Data streaming services to Vehicular Clients (VCs) must cope with delay and delay-jitter issues, while mining large volume of data. To this purpose, Fog computing (FC) is an emerging and still unexplored paradigm that aims at distributing small-size self-powered networked data centers (e.g., Fog nodes) between resource-rich remote Clouds and resource-limited smartphone-equipped VCs, in order to perform context/content aware energy-efficient data mining and dissemination.

Overall, main goals of the V-Fog2 project are to complete the design and validate in SW the related algorithms already developed during the FIRST year, specifically:
a) the distributed and adaptive real-time algorithms for the QoS mining of environmental multimedia Big Data streams;
b) the minimum-bandwidth algorithms for the QoS energy-efficient migration of Virtual Machines (VMs) to/from Fog nodes over multipath-TCP connections;
c) the adaptive Machine Learning-based mining algorithms for the real-time context-aware detection and classification of sensor-acquired human activities of specific social interest.