Fog for 5G and IoT

10 Distributed Online Learning and Stream Processing for a Smarter Planet

DEEPAK S. TURAGA¹ and MIHAELA VAN DER SCHAAR²

¹ IBM T. J. Watson Research Center, Yorktown, New York, NY, USA

² Electrical Engineering Department, University of California at Los Angeles, Los Angeles, CA, USA

10.1 INTRODUCTION: SMARTER PLANET

With the world becoming ever more instrumented and connected, we are at the cusp of realizing a Smarter Planet [1], where insights drawn from data sources are used to adapt our environment and how we interact with it and with each other. This will enable a range of new services that make it easier for us to work, travel, consume, collaborate, communicate, play, entertain, and even be provided with care.

Consider the pervasiveness of the mobile phone. It is rapidly emerging as the primary digital device of our times—with over 6 (out of the 7) billion people in the world having access to a mobile phone [2]. We are witnessing the rapid emergence of services that use these phones (especially smartphones) as sensors of the environment and interfaces to people. For instance, it is now common with several map services (e.g., Google Maps) to be provided a live view of the traffic across a road network. This aggregate view is computed by processing and analyzing in real time the spatiotemporal properties of data collected from several millions of mobile phones. Applications such as Waze include adding crowd‐sourced information to such data, where individual people use ...