138 W.P. Worzel
Internet. City infrastructure and buildings have an abundance of sensors as well,
including water sensors, traffic sensors, and environmental sensors. As the number
and variety of devices increases, it creates an increasing computational burden to
filter and integrate data in real time. The estimated level of 20 billion new devices
will make the model of sending raw data for processing unsustainable both in terms
of the communications infrastructure and the raw processing power needed at data
centers. In particular, finding, collecting and analyzing data in real time from this
many devices becomes an impossible task. Instead it will be necessary for some
level of data processing to be done at the edge and the results will be integrated
locally and shared globally. Cloud Computing assumes that bulk of computing
done will in ‘The Cloud’ which usually means at distant data servers such as those
provided by Amazon, Google and other providers of computing resources. Most
uses of the Internet of Things rely on The Cloud to process data generated at
the edge. In contrast to Cloud Computing, the term ‘Fog Computing’ is used to
describe a more diffuse form of computing that is closer to the ground and where
analytics is done locally and potentially shared with many users. This approach
is already becoming an important part of the discussion. Cisco has introduced an
operating system called IOx in Cisco (2014a) that is already being used in industrial
applications, including in Smart Grid applications described in Cisco (2014b). The
AllSeen Alliance ( 2012 ) has created an open source OS called AllJoyn that is
particularly useful for in-home devices. Meanwhile Google has recently announced
a version of its Android OS called KitKat described in Google ( 2015 ) that is
optimized for low memory devices such as is common in Edge Computing. But
while these advances are important, they focus only on the local aspects of Fog
Computing rather than a vision that integrates the Fog with the Cloud; it is assumed
that the Cloud as it exists currently is all that is necessary to take advantage of the
coming data explosion in the Fog.
2 Fog Lifter
TMThe author has been working collaboratively on a suite of open source software that
will facilitate the full integration of Fog Computing into the Internet. Fog Lifter is
built on the assumption that local computation and results will be used for multiple
purposes, allowing geographically dispersed information to be combined with other
data for very different purposes. This may be summarized as “Compute locally,
analyze anywhere” where the results of local computation is made available through
a registry and may be used as part of a larger computation, whether in the Cloud or
at another ‘locality’.