A convergence of technologies has created the notion of the Internet of Things (IoT) to rev-
revolutionize areas such as home automation, environmental monitoring, hazard detection and
industrial IoT applications. A major barrier to the pervasive deployment of IoT technologies
is the availability of energy. Cisco predicts that there will be 50 billion devices by 2020 [1],
powering this number of sensors on primary batteries creates a momentous task when they need
to be replaced and there is also the subsequent environmental concerns of battery disposal.
The need for a deploy and forget IoT solution has driven research into energy in Wireless Sensor
Networks (WSN) over the past decade. The current state of the art, Energy Neutral Operation
(ENO), utilizes techniques such as sample rate control, adaptive duty cycling and data reduction
methods to balance energy generation, storage and consumption. However, despite considerable
work being undertaken in this field, a systematic literature review undertaken for this thesis
has highlighted gaps which remain in the state of the art. From this literature review, three
keys questions are highlighted for exploration:
• Energy generation, although variable, is not controllable or optimizable in-situ. If ENO
WSN could dynamically adjust their energy generation to optimize the energy harvesting
methods through actuation, would this produce greater energy for sensor systems? A
lightweight distributed energy generation optimization algorithm has been developed for
solar tracking which produced an average of 9.2% increase in energy generation across all
testing.
• Energy consumption in ENO WSN is controlled via adaptive duty cycling. Tasks
are scheduled between a predetermined minimum and maximum duty cycle within an
individual time-slot to ensure less energy is consumed than is generated. However, this
adaptation is done only with consideration of the energy availability and without any
knowledge of how the downstream information is used by applications. Could a network
disseminate user requirements to nodes to better provision tasks in a network? Would
this be more efficient and could this free up energy to be utilized for other tasks? To
answer these questions, an application and energy aware tasks scheduler has been created.
In particular, the Surplus Energy Task (SET) scheduler improves energy utilization by
an average of 21% .Battery Health There is the proposition that WSN can now be deployed and operated
in perpetuity with the evolution of energy harvesting methods. However, the nature of
energy storage is such that batteries degrade over time to the point they can no longer
hold charge. Can ENO WSN systems be made aware of battery degradation in node and
then use this information as a parameter to be optimized to promote longevity in ENO
WSN systems? In answering this question, methods created resulted in an average of
205.2% improvement in lifetime without a reduction in average performance of a sensor
system across different energy harvesting source tests.
In this thesis, methods, models, and algorithms are put forth to answer these questions. The
themes and methods discussed are of particular interest to researchers in the areas of Internet
of Things (IoT), Wireless Sensor Networks (WSN), Cyber-Physical Systems (CPS), Energy
Neutral Operation (ENO), and Information and Communications Technology (ICT) for the
promotion of sustainability in future cities.