Page 45 - European Energy Innovation - Spring 2016 publication
P. 45

Spring 2016 European Energy Innovation         45


use of both local smart application
gateways and cloud computing.
COGNICOM shares some common
features with Fog computing in
terms of bringing computing closer
to the end user, however with one
key difference: where computation
occurs will be decided by the CE
to maximize utility, reliability, and
privacy and minimize latency and
energy expenditures of the entire IoT
networks. At the heart of the CE are
deep-learning algorithms organically
integrated with advanced game-
theoretic decision analytics to supply
cognitive functions for selective smart
objects as well as the complete IoT
application. An equally important
feature of COGNICOM is Smart
Connectivity (SC), which enables
seamless, energy-efficient and reliable
connection to the cloud, smart-objects
and other IoT devices and sensors.

3 COGNICOM CONCEPT                        Fig. 2: COGNICOM Implementation based on SAG.
The COGNICOM architecture depicted
in Fig. 1 consists of two entities : the  and context awareness to be exploited    SC will leverage the CE to support
CE and the SC. The CE concept is an       in future orientation/reasoning phases.  Dynamic Spectrum Access (DSA). SC is
extension of the previous research        It should be powerful enough to enrich   multifunctional, flexible and scalable.
on cognitive radio8,9. CE is defined      the knowledge base, to foster the
as an intelligent agent that manages      increased efficiency of the reasoning    4 COGNICOM IMPLEMENTATION
the cognition tasks. First, CE performs   and to enhance the decision. As a        The IoT hybrid architecture shown in
observations, collects data and extracts  result, there is a close interaction     Fig. 2 is inspired by a past trend in
the information and the knowledge         between learning, knowledge,             mobile communication, where base
regarding environment or user. The        reasoning and decision, which            stations became smaller, less expensive
CE then reasons in order to analyse       complement each other to improve the     and more capable over time (micro-
and classify the situation and decides    operation of the system as a whole.      cells, pico-cells and femto-cells10).
on the appropriate response. Once                                                  The idea is to move away from cloud
decisions are made, the CE adapts         The SC enables connections with the      computing and Fog computing, and
and reconfigures its parameters with      cloud and other objects to amplify the   to leverage local computing whenever
respect to user-defined objectives. The   capabilities and the value of the CE.    possible. This not only reduces costs,
learning component of CE evaluates        Its goal is to enable connectivity to    boosts capacity, reduces latency and
the outcomes of the decisions and is      every device everywhere and anytime.     speedups network expansion, but
responsible for building up knowledge
   40   41   42   43   44   45   46   47   48   49   50