Power Supply Dynamics for Outdoor IoT Sensors in New Zealand

Abstract

The paper documents the authors' power usage experience and hardware related designs, and concludes with a summary of the leading options available, with indications of the factors to be considered in choosing the options to use for a particular situation. The authors are committed to open standards-based, accessible, and affordable environmental monitoring, as part of national initiatives to restore biodiversity and plan for climate change. Research is presented using an Iterative Action Research model. Using a formal model allows the authors to highlight the common themes in their research while contextualising differences in the way they approach a common challenge. Power requirements are critical to a successful IoT deployment in any environment which doesn't have a continuous power grid available. Following the guidelines presented in this paper will enable a sound platform for building a successful IoT environment.

Full text

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Power Supply Dynamics for Outdoor IoT Sensors in
New Zealand
Steve Cosgrove
Whitireia New Zealand
steve.cosgrove@whitireia.ac.nz
Matthew Way, Ann Way
Econode New Zealand
matt@econode.nz, ann@econode.nz
Andrew Hornblow
Educational Consultant – Internet of Things
80 King Street, Opunake
andrew.hornblow@gmail.com
Abstract
The paper documents the authors’ power usage experience and hardware related
designs, and concludes with a summary of the leading options available, with indications
of the factors to be considered in choosing the options to use for a particular situation.
The authors are committed to open standards-based, accessible, and affordable
environmental monitoring, as part of national initiatives to restore biodiversity and plan
for climate change.
Research is presented using an Iterative Action Research model. Using a formal model
allows the authors to highlight the common themes in their research while
contextualising differences in the way they approach a common challenge.
Power requirements are critical to a successful IoT deployment in any environment
which doesn’t have a continuous power grid available. Following the guidelines
presented in this paper will enable a sound platform for building a successful IoT
environment.
Keywords
Power Consumption, Internet of Things (IoT), Solar, Open Source, Iterative Action
Research
1. Introduction
The term ‘Internet of Things’ (IoT) describes a growing network of Internet connected
devices of various types, used in all areas of 21st century life. Many of these devices
are sensors, predominantly commercial products that are used by the million, in
situations like car parking or street lighting (IEC, 2016).
This paper considers power requirements of those sensors that are in challenging
situations in New Zealand. In particular, they might be in locations that are special by

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
being remote, sensitive, out of digital communication range, in harsh environmental
conditions or similar circumstances. These can be on small islands or challenging
situations on the mainland islands.
being remote, sensitive, out of digital communication range, in harsh environmental
conditions or similar circumstances. These can be on small islands or challenging
situations on the mainland islands.
being remote, sensitive, out of digital communication range, in harsh environmental
conditions or similar circumstances. These can be on small islands or challenging
situations on the mainland islands.
In this section we introduce the conservation environment of New Zealand, to show the
place of this paper within the broader picture.
1.1 Structure of Organisations in Conservation
Reports in the media and relevant websites suggests New Zealanders are becoming
increasingly engaged in conservation. Craig et al. (2013) conducted an early analysis
which was widely reported in various fora. This section cites a number of organisations
that have been established since 2013.
It is a very complex domain. There are numerous organisations involved with various
inter-relationships. The authors find it useful to use a triangle model to describe different
views of these inter-relationships. The metaphor of using the mechanical strength of the
triangle to describe many aspects of human society is eloquently expressed by
Kim et al. (2001).
The most significant driver of New Zealand conservation is removal of predators.
Towns et al. (2001) show how one range of native creatures (reptiles) can thrive when
introduced predator numbers are reduced. They looked at the end of the 20th century.
Twenty years later this strategy has caught the imagination of the population,
significantly boosted by a Government-led campaign with the stated aim to remove all
introduced predators by the year 2050. The Predator Free NZ Trust (PFNZ) lists over
twenty ‘key players’ (https://predatorfreenz.org/big-picture/pf-2050-vision/). Many of
these groups have thousands of members.
One way of viewing the ‘Big Picture’ would be a triangle consisting of three interrelated
drivers:
• Funding: Dominated by Government body Predator Free 2050 Limited (PF2050)
and the private NEXT Foundation, along with thousands of smaller players
contributing money, capital and labour. (https://pf2050.co.nz/
https://www.nextfoundation.org.nz/)
• Community Efforts and Knowledge: Organisation supported by PFNZ,
including Iwi, community groups, landowners and others, many identified in the

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
PFNZ link above. One of the current authors – Andrew Hornblow – is
represented here.
• The Research – Technology – Product cycle: Some large organisations active
here include Zero Invasive Predators (ZIP), Department of Conservation and
Operational Solutions for Primary Industries (OSPRI) (https://zip.org.nz/
https://www.doc.govt.nz/ https://www.ospri.co.nz/about-us/). The education sector
has a key role, and there are private companies contributing in various ways.
Two authors, from Econode, are in this area (https://www.econode.nz/aboutus).
These three drivers support each other in a virtuous cycle. The Community has been
undertaking conservation work that can be traced to the 19th century
(https://www.forestandbird.org.nz/about-us/our-history). That long-term background has
contributed to a country that creates and encourages funders. With funding, the
Research community can create new products which extend what the community can
do, and cycle continues.
1.1 The need for sensors
The community driver outlined above has achieved significant results over the past
hundred years using very effective devices created from simple ‘seat of your pants’
technology and knowledge. Grant Ryan of the Cacophony Project
(https://cacophony.org.nz/) makes a case that Information Technology could improve the
efficacy of predator control by a factor of 80,000 times (TEDx Talks, 2016).
There have already been significant examples of regeneration and recovery of native
flora and fauna through the work of Predator Free 2050 Limited (PF2050, 2020).
Sensors are important in both the predator removal process and monitoring of species
recovery. One of the early documented accounts of sensor use in the type of
environment considered in this paper is ZIP (https://zip.org.nz/). This organisation
contributed to development of the concept of a ‘virtual fence’ and documented progress,
including the use of sensors, in their inaugural ‘annual’ report (ZIP Report to 30 June
2015, 2015)
1.2 The place of Electric Power Dynamics in Conservation
An analysis and prediction of ‘Number of Internet of Things (IoT) connected devices
worldwide from 2019 to 2030, by vertical’ asserts there were about 7.8 billion devices in
2020, which is predicted to increase to about 22 billion by 2030 (Transforma Insights,
2020). The vertical market ‘Agriculture, Forestry & Fishing’ is the closest fit to
conservation. This market is predicted to grow to about 20 times its current size. This is
a predicted growth rate averaging seven times the growth rate of other markets.
The authors believe that sensors will proliferate among those 22 billion devices,
however conservation will have the greatest reliance on use of sensors which lack an
established power source.
A case can be made that power dynamics for low-power sensors will be a significant
influence on conservation work over the ten years. Our analysis of the literature

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
suggests this is an area that is currently underreported. The underlying electrical
principals date back to the 19th century. Some older literature such as Belhadj-Yahya
(2010) can inform modern usage, but the way that electricity is generated and used
continues to improve at a great rate. Battery selection can be informed by publically
available datasets such as dos Reis et al. (2021), but battery technology is just one part
of the power dynamic question. This paper will consider a broad range of options,
informed by the authors’ experiences.
We suggest that makers of products supporting New Zealand predator eradication can
be modelled on a matrix of two axes:
• Commercial Imperative represents the degree to which the product producer
needs to make a monetary profit or surplus.
This will clearly affect the investment choices. A producer without a strong
commercial imperative might have access to non-monetary resources, particularly
labour.
• Design Kaupapa represents the approach the producer has to making design
elements available to the community.
There is a long history in the literature documenting the benefits of open sourcing
of product intellectual property, particularly to those without financial resources
(Lerner & Tirole, 2005; Watanabe, 2001). Grant Ryan – Cacophony Project
initiator and coordinator, articulates this position well (TEDx Talks, 2016). There
could be value in more organisations being placed appropriately on Figure 1 A
Matrix of Design & Commercial Dimensions. The example below has the authors’
products positioned on the matrix for further reference in this paper.
•
high Commercial imperative low
Closed Design Kaupapa open

Econode


Andrew Hornblow

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Figure 1 A Matrix of Design & Commercial Dimensions
1.3 Structure of this paper
This paper starts with an introduction to the subject matter. This includes a context-
setting description of the New Zealand conservation industry and the place of electric
power dynamics in sensor networks.
The academic research method is described, including how the authors see it applied in
the context of the subject at hand.
Three subsequent sections describe the options available and how each of the authors
apply those options in their work. Application is described in terms of the previously
defined research method.
Finally the paper concludes with suggestions for further research and a summary of how
the authors see the academic community making a strong contribution to the future of
New Zealand biodiversity.
2. Aims and Research Design
This paper will contribute to the readers’ understanding of parameters of the New
Zealand conservation sector. That understanding will inform identification of a set of
parameters to use when provisioning power to sensor deployments in an efficient and
effective manner. The result will include open-source design guidelines that the
research community can implement and further analyse for the greater good (Lerner &
Tirole, 2005).
We will answer the question: “How can the most effective power supply solution be
determined for a sensor in any specified environment?”
The literature shows solutions that were applicable in the 20th and early 21st century
(Belhadj-Yahya, 2010). In the past ten years all technologies in this field have been
significantly improved and optimized, while prices have fallen. This project builds on
previous work to develop new designs using current and emerging technologies
specifically intended for power-challenging situations in New Zealand. The authors will
then implement and analyse the outcomes to create a source of current information.
2.1 Research Method
The research uses an iterative action research method to build this contribution to the
wireless sensor literature. This concept of adding iteration to an active research model
was introduced by Susman & Evered (1978). The authors have reduced Susman &
Evered’s five phases to four as shown in Figure 2. The iterative approach to action
research was well articulated and supported by Kock et al. (1997).

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Figure 2 A cyclical process of action research adapted from (Susman & Evered, 1978)
These phases very closely map the way the authors of this paper structure their work:
• Diagnosing – This paper has established a very broad picture in section 1
above. The detail of problem definition within that broad picture is based on
various criteria discussed in this paper.
• Action – The authors have extensive experience in their domain. They are able
to take the problem, identify and evaluate potential solutions, then construct an
artefact that is expected to solve the problem. (Combines Susman & Evered’s
‘Action Planning’ and ‘Action Taking’ phases.)
• Evaluating – Studying the consequences of implementation of the artefact.
• Specifying Learning – It is essential that consequences are evaluated and
articulated in an appropriate form to continue the cycle of diagnosis and a new
iteration.
2.2 Applying Iterative Action Research
Collectively, the authors have experience across many sectors of the IoT
community. This includes: commercial sensor deployment across a broad range of
remote and urban environments; controlled and open field experiments; desk-based
analysis; all using a wide range of sensor hardware characteristics. Over several years
those characteristics have been modified for each application and to accommodate the
needs of each location. This research design has resulted in successive iterations in the
evolution of sensor power supplies.
Our non-commercial author, Andrew, started his work in this area by defining a very
general problem – School children are not being sufficiently engaged in science.
Diagnosing –Identifying
or defining a problem
Action –Considering
alternative courses of
action & selecting one for
solving a problem
Evaluating –Studying the
consequences of an
action
Specifying Learning –
Identifying general
findings

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Hipkins (2009) is one of many authors who have articulated various aspects of this
problem. This paper considers Andrew’s research with power dynamics for IoT sensors.
In that context, he has implemented the Action Research process relying on his own
resources which are strongly electronics and hardware focused. His diagnosis of the
power problem, and planning to act on that diagnosis generally relies on his own
experience. The action phase tends to focus on smarter use of currently available
resources.
Econode has been built around developing open and commercial solutions to a
particular big challenge – The New Zealand Pest Problem (Way & Way, n.d.). Within
that challenge they diagnose the current state of the industry to identify a particular
problem. That problem can come from any stage of the product life cycle. For
Econode, the action phase requires consideration of a very broad range of options.
Appropriate technologies are acquired or developed which are used to develop a
solution to a particular problem, while being mindful of the interaction of that solution
with the entire Econode infrastructure.
The authors suggest that it is in the evaluation and specification of learning phases that
their different kaupapa results in quite different needs and actions.
After introducing a new solution, Andrew gathers his own observations together with
those of his research collaborators, informed by real-time data using the Cayenne (n.d.)
IoT visualisation platform (http://bit.ly/Penguins2021). He then discusses these
observations with his collaborators, with the help of annotated diagrams such as Figure
3 Example of Annotated Penguin Sensor Data. The evaluating phase uses this data to
identify the consequences from each iteration. Things learned from the iteration are
then used to identify another problem to work on. The action research cycle then
continues.
For problems in the context of power dynamics, Andrew has identified that the most
effective action he can take is often reducing the number of parts, or finding ways to
reduce the power consumption of the parts being used. This is consistent with the
observation above that Andrew relies to an extent on using his own electronics and
hardware resources. On the other hand, when an appropriate new technology becomes
available, the non-commercial product development model gives Andrew agility to be
able to take advantage of that technology quickly. Such an agile model requires
participation from all stakeholders. While this works for many partners, largely in the
conservation and school sectors, other people in these sectors prefer to work with a
science partner who has a different business model. This results in Andrew’s
observation “Making things simply using the least parts and energy makes no-one
famous or rich.”

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Figure 3 Example of Annotated Penguin Sensor Data
Econode conducts the ‘Evaluating’ phase thorough evaluating, feedback and specific
learning from a broad range of projects. Each solution has been targeted at a particular
predator free project. They evaluate what's working and what should be improved. This
is informed by input from contractors who install their products in the field, and by
monitoring the product's performance, which is assessed during each cycle. After
identifying what has been learned, new problems can be identified, and the cycle
continues.
This process has enabled Econode to enhance the product's performance by taking
advantage of the current availability of technology in the market. When they started
using LoRaWan and IoT technology in 2015, there were not many support devices
available. They had to build their own complete solution. Now there are many options
in the market and development can focus on the flagship product, the SmartTrap. This
process has enabled Econode to add significant improvement in both design and
functions to each generation of SmartTrap nodes.
3. Options for Power in Isolated Locations
Having previously defined the scope of this paper, this section describes various power
supply paradigms and devices, then considers problems and some solutions. Finally
longer-term issues are described. The following section describes some case studies
using the options discussed here.
There is a very wide range of options being developed which challenge long-held views
of how power can be provisioned.
Note that while technically a battery has been defined as a group of two or more cells
that furnish electric current, common usage sees the term also used to describe a single
cell (Merriam Webster, n.d.). This paper will use the term ‘battery’ to refer to one or
more cells.

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
3.1 Power Storage
Batteries have been used for remote power supplies since the 19th century
(Whittingham, 2012). The authors see this continuing, however the batteries continue to
change. At a basic level, a battery is described in terms of the two metals (an anode
and a cathode) that are the active components. Since 1991, the anode in most
rechargeable batteries has been Lithium. A variety of metals have been used as the
cathode, and alternatives to Lithium are also being developed. It seems likely that
different applications will better suit one battery technology over another (Placke et al.,
2017) into the foreseeable future.
The authors divide sensor battery needs into two major categories: Primary batteries are
those used for small sensor devices, generally using a very low power ‘System On a
Chip’ (SoC); Secondary batteries power infrastructure devices that have higher power
requirements.
Three factors are converging to lower the cost of primary batteries: devices are using
less power – now measured in a decreasing number of microamps (µAmps): battery
performance is improving, both the active capacity of the battery and the expected shelf
life; prices are dropping. The authors consider Lithium-Thionyl (Li-Socl2) as the highest-
performing option for a primary battery. Unfortunately, they are also very expensive, in
the order of $20 for one AA cell. On the other hand AA alkaline batteries are cheap,
simple, and reliable. The authors have used Eclipse AA batteries 40 of which can be
purchased for under $15 (Jaycar, n.d.). These can be relied on to have a shelf life of
five years and active capacity to power a simple sensor for that time.
Sensors designs now allow for very low power demand, which means less power is
required to run them. Single use primary batteries should power a sensor for many
years.
Secondary batteries need to be very carefully considered. These power a very wide
range of devices, typically measured in hundreds of milliamps – orders of magnitude
more that primary devices.
The authors have found that Li-Socl2 batteries are significantly more reliable. They
have excellent shelf live and retain full capacity for longer than other chemistries. This
needs to be weighed against the high price. Another option that has been found to be a
reliable secondary power is a lithium-ion polymer battery (LiPo).
Another power storage option is a Super-Capacitor. These can be compared with a
rechargeable battery but are capable of a virtually infinite recharge cycle.
Whittingham (2012) describes uses for these in situations where there is a significant
amount of space to hold the capacitor. The authors agree with Whittingham that these
will be considered alongside batteries at some point in the future. The technology is not
yet suitable for small, portable, applications.

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Considering these notes in the context of Figure 1 A Matrix of Design & Commercial
Dimensions, while the authors agree on the characteristics of power storage
technologies, their individual dimensions result in different choices.
Econode are developing commercial solutions for sites that can be very expensive to
access. For primary storage, their designs currently use easily replaceable batteries
that can be relied on for five years or more. Often AA size Alkaline batteries are
sufficient here. In some circumstances the long life of Li-Socl2 will be worth the
significant extra cost. Secondary storage tends to high quality PV panels with non-
replaceable rechargeable batteries appropriate to the site. Econode is developing new
technologies with ever decreasing power usage. As relay and gateway (secondary)
devices get more efficient, they might also move to primary battery types.
Andrew’s costs are not always easily recovered, and his client base is more diverse.
His activities can start with a pre-school child, a button battery and LED light, through to
a device to read and remotely report the Radio Frequency IDentification (RFID) tag on a
passing penguin (Mattern & Seed, 2020, p. 17). This leads to a very wide range of
power storage choices, ranging from AA alkaline batteries on devices used for
demonstration and short-term/low current projects, while generally using a ‘happy mix’ of
any PV panel matched with a permanent LiPo battery on medium to long term projects.
3.2 Power harvesting
We established in section 3.1 above that the authors have a need to recharge batteries
for secondary power requirements. In the context of this paper, a harvesting solution
must be able to generate energy in isolated locations.
Secondary power is typically required by a wireless repeater, which often picks up
signals from a sensor (running on primary, non-rechargeable power for multiple years)
and repeats that signal so the signal can be received by a gateway device which will
transfer the signal to a suitable cloud provider. Repeaters can also be configured to
repeat signals from each other. The gateway device does, itself, consume secondary
power, at a higher rate than a repeater.
Solar photovoltaic (PV) panels are very commonly used to charge the secondary battery
discussed in section 3.1 above. Modern monocrystalline PV cells have made panels
smaller and more efficient (in the order of 15-20%) than polycrystalline ones previously
used extensively (13-16%). Econode has observed this gain in PV efficiency, coupled
with reduced power usage and more efficient power monitoring and management
(results from the iterative research process). The process lead to Econode having a
design goal to create a versatile repeater with smaller physical size meaning less weight
and wind load. This is described in section 5.2. Another important component of the
Econode design is to have the batteries and electronics directly under the solar panel,
so the panel forms a roof.
An air gap between the solar panel and electronic casing will provide cooling for the
batteries, lithium batteries don't handle heat well.

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Vibrations contain energy. Research in this field tends to be concentrated on
opportunities to harvest energy created by the movement of motor vehicles.
Okkeh et al. (2018) suggest that street design can service this purpose. Vehicle drivers
would likely agree with Wang et al. (2016) that there is a significant energy exchange
when a vehicle goes over a speed bump. None of the literature found, or experience of
the authors, has suggested a solution for energy harvesting from vibration in remote
areas yet, however it is a developing field.
The remote areas of New Zealand often have significant qualities of wind and water.
These are both excellent sources of energy. The experience of the authors, however,
suggests there are significant problems with harvesting this energy for very small scale
projects.
The reason for sensors being in remote areas is generally associated with retention of
the natural environment. Robust and sustainable water-based electricity generation
tends to conflict with that objective. Less invasive methods tend to be prone to failure
when subjected to natural threats such as flood.
Wind harvesting has been used to generate electricity in various situation for many
years. Once again, in the locations being targeted by this paper, the traditional wind
turbine is often too bulky or intrusive to be a viable solution. Rotating blades also
introduce the hazard of bird-strike (Straka et al., 2020). This risk is unacceptable where
there are vulnerable bird species present. One wind technology that has some potential
is piezoelectric electricity generation. This process is well described by
Wang et al. (2019). They did not, however, use the electricity generated to power
sensitive electronic circuits of the type commonly use it in a sensor infrastructure.
There are a number of other options being researched or used in limited areas. An
example is Wang et al. (2019), mentioned in the previous paragraph. Other harvesting
techniques, “both well-known and novel”, are explored by Curry and Harris (2019). One
potential future energy source they mention is generating electricity from biological
sources. This research area is further developed by Ayala-Ruiz et al. (2019). Similarly,
Elahi et al. (2020) review “energy harvesters based on mechanical, aeroelastic, wind,
solar, radiofrequency, and pyroelectric mechanisms”.
3.3 Common Issues for PV Panels
The most obvious question when considering power dynamics for sensors is current
draw of each device being used. This is an area where the authors take different steps
to achieve the same outcome, as described in section 2.2. Andrew works on achieving
savings through clever use of software and different components that can result in
power consumption savings. Econode use a more data-intensive approach to monitor
device performance, and aggressively research new technologies for savings.
Specifications for a particular application need to be carefully thought through and
planned. A person requesting a sensor may not understand that a remote radio data
link can easily account for 90% or more of energy requirements. Simplifying the data

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
collection cycle by reducing the number of readings can make a significant difference to
overall energy consumption. An example Andrew has found is that ambient temperature
only needs a once in 10 minutes duty cycle and rounding the value being transmitted
can save much radio data. ‘Smart’ pre-processing is essential. Smart choice of what
works saves cost and complexity.
The most commonly used energy harvesting techniques listed in section 3.2, wind,
water and sun, each have issues identified in that section. The common factor is a need
to alter the environment to accommodate each technology. In the context of this paper,
the solar photovoltaic (PV) panel is the significant device for provision of electricity.
While the research process generates new solutions, traditional PV panels will remain in
widespread use. These can be very site specific if they need to supply a significant,
consistent power. Where there is a good understanding of the power requirements,
Andrew has found that a lithium-ion polymer battery (LiPo) battery and suitable size PV
panel can be a simple and effective solution.
The biggest challenge is external material that prevents sun getting to the panel. Some
of the common examples in conservation work are:
• Guano from animals, birds, reptiles and insects.
• Dust, dirt and leaves blown or fallen on to the panel.
• Growth of plants around the panel (for medium to long-term installation).
• Overhead canopy of trees shading the sensor. This might not be obvious during
installation. The sensor in Figure 4, with a small PV inside the case appears to
be in clear space but delivered poor results for most of the day. The same
sensor worked flawlessly when in clear sun. The larger PV shown at the same
location in Figure 5 worked well.
• Disturbance by passing creatures. Generally this can be predicted, and the PV
panel secured to mitigate this risk. An unexpected disturbance is shown in Figure
6, where it appears a passing Little Spotted Kiwi sensed insects or invertebrates
moving under the sensor and tipped it over so it could eat them.
3.4 Resilience and Digital Twin
Econode is integrating the concept of digital twin in new developments. The term can
be defined as “a virtual representation of a physical asset enabled through data and
simulators for real-time prediction, optimization, monitoring, controlling, and improved
decision making.” (Rasheed et al., 2020).
This will be achieved by inclusion of a high precision volt and current meter ( +/- 2mv ),
so Econode will be able to get a lot of statistical data about the panel performance. This
is being achieved by using two INA226 modules (https://www.ti.com/product/INA226),
one on the solar panel, and a second one on the batteries. Along with regular sensor
data, these power statistics will be sent to the IoT cloud provider.

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
The purpose of the Digital Twin is to monitor the device under test over its operational
life time. One outcome could be should foliage grow over the solar panel, or a solar
panel gets bumped or knocked, this will show as an anomaly and action taken. As the
Digital Twin is informed by monitoring, environmental changes over time can be
predicted, and combined with estimated run time on the internal batteries.
Figure 4 Sensor with small PV at
ZEALANDIA
Figure 5 Sensor with large PV at ZEALANDIA
Figure 6 Sensor tipped over by a Kiwi in
ZEALANDIA

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
4. Case Study – Andrew Hornblow
The authors’ approach to iterative active research was outlined in Section 2. Aims and
Research Design. This section summaries some examples of the work of one author,
showing how he applies the research design process
4.1 Taranaki Penguins
Andrew’s longest-running project has been thermal detection of Penguin nest box
occupation around the Taranaki coastline (Figure 7). Initially these used simple data
logging devices which needed to be manually read periodically. The iterative research
process saw the development of sensors with a remote wireless link to a locally hosted
IoT backend of the time.
For the first five years three AA alkaline batteries would last a breeding season of about
6 months. Battery life decreased as more data was being transferred. A recent
development is use of 250mAh LiPO batteries charged by a small solar PV. The
evaluation phases of this change established that excess charge is extremely unlikely to
be an issue. Power is managed by balancing load to state of LiPO charge. When there
are sustained periods of sunshine, the transmitters are left running for longer to avoid
overcharging.
4.2 Schools Digital Curriculum
During the years 2016 – 2018, The New Zealand Ministry of Education (MinEdu) ran a
pilot Digital Technology learning unit. As part of that, Andrew contributed to a project
where students built data loggers based around an educational microprocessor, the
PICAXE 08M2 (Revolution, 2021).
Students built a device to monitor a range of environmental and building metrics. Those
metrics included soil dampness, water table levels, dwelling temperature, bedroom
timber moisture and temperatures. The program was very short but the simplicity of
single processor and 3 AA Alkaline batteries, thermistor or nail moisture probes meant a
that the cost to a school was less than $10 per student. This could be expanded by the
addition of a basic Message Queuing Telemetry Transport (MQTT) broker that
connected these to a free IoT cloud service (Cayenne, n.d.) using the MQTT TCP/IP
protocol. The broker ran on a Raspberry Pi computer on the school Local Area Network
(LAN). Andrew asserts that MinEdu took some interest in the building timber moisture
testing possibilities!
One secondary school developed this theme, and engaged with Andrew to produce a
significant ‘IoT Course’ which enabled students to work on unit standards in three parts
of the Digital Curriculum (Wanaka Sun, 2017, p. 7).
• Students completed an electronics unit while assembling the sensors.
• Programming the sensor completed a Computer Programming unit.
• Creating a ‘dashboard’ to display results completed a web page development
unit.

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Figure 7 Penguin nest boxes near the Port of Taranaki
4.3 Taranaki Sound Lures
Primary aged students at Auroa School in South Taranaki identified that the whio duck
was endangered by predation from stoats in the streams on their farms. They worked
with Predator Free Taranaki (https://predatorfreenz.org/5126-2/) and local existing
farmer pest control groups. They were supported with electronics support from Andrew,
education guidance from the school deputy principal and conservation support from the
local Rotokare Sanctuary and the Department of Conservation.
The project generated considerable interest. Students and the school have been
acknowledged by the South Taranaki Regional Council and the annual Tahi Rua Toru
Tech! national school technology competition (https://123tech.nz/) where they won their
grade in 2020 (Groenestein, 2021).
The project required assembling and programming a PICAXE-based circuit that included
a module holding an MP3 player and 3 Watt audio player. The code developed by these
students selected a sound track that could attract the target species while manipulating
volume and using solar load management for optimum battery life (Taranaki Mounga,
2020). In May 2021 the first unit of this design to be deployed was retrieved for
maintenance. After two years literally ‘in the field’ running with a medium sized PV
panel and a 2000mAh LiPO battery, it was found to be working well, just needing the
replacement of some rusty terminal screws and bolts. Andrew sees this project as an
example of the success of the iterative action research cycle. The next problem
identified has been lack of feedback from this device. The project team aim to use a full-

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
duplex LoRA connection to their IoT cloud to control experiments in real time.
Understanding the enemy and closing the loop on what works!
5. Case Study – Econode
It was made clear in Section 2 that the authors who create products apply the iterative
action research methodology in different ways. Here the Econode authors outline an
example of their data-centric approach to the research design process.
5.1 Battery Use in Econode nodes
Software developed by Econode records a range of telemetry data from each node,
some of which is shown in Figure 8. This enables monitoring and analysis of the node
operation.
This was used in the ‘Gen5’ nodes deployed in November 2019 for Predator Free
Taranaki. Each node is powered by four consumer-grade alkaline batteries (Jaycar,
n.d.). At the time of writing, the highest price for these batteries was about 42 cents
each, or $1.70 for a set of 4. This is an example of the Econode kaupapa of keeping
costs down contrasts with another sensor supplier who list a replacement battery pack,
with a comparable duty timeframe, for $50 (http://bit.ly/OutPostInfo).
Firmware used in the Gen5 nodes used in this trial require 3.3 volts. They use 4 μA in
sleep and about 70 mA when transmitting. Each two hourly transmission takes about
2 seconds. Battery voltage decline is shown Table 1 Actual power consumption
Gen5Rev3 SmartTrap nodes PF2050 Taranaki. An example of an Econode circuit
board is provided in Figure 9.
Table 1 Actual power consumption Gen5Rev3 SmartTrap nodes PF2050 Taranaki
4 x AA alkaline batteries
Date:
Node
2019-
11
201
9-12
202
0-01
202
0-02
202
0-03
202
0-04
202
0-05
202
0-06
202
0-07
202
0-08
202
0-09
202
0-10
202
0-11
202
0-12
202
1-01
202
1-02
202
1-03
202
1-04
202
1-05
4023
6.33
6.26
6.23
6.11
6.07
6.05
6.04
6.01
6
5.99
5.95
5.94
5.84
5.82
5.81
5.75
5.53
5.48
5.46
4029
6.25
6.24
6.22
6.2
6.06
5.95
5.84
5.77
5.72
5.71
5.66
5.67
5.64
5.63
5.6
5.59
5.57
5.52
5.47
4030
6.5
6.42
6.22
6.12
5.99
5.93
5.9
5.87
5.8
5.78
5.77
5.77
5.77
5.76
5.75
5.74
5.75
5.74
5.74
These figures suggest that about two and a half years battery life is achievable.
Econode estimates that standard lithium batteries would power a node for about five
years with a price of about $20 for a pack of four. It is likely that a pack of lithium-thionyl
chloride batteries would have double the lifetime – about ten years.
5.2 Secondary Battery use in Econode Infrastructure
As described in section 3.1, the authors distinguish primary battery needs (as described
in the previous section) and secondary battery needs, for infrastructure components that
have a higher current drain. This section introduces power needs for LoRaWAN radio
gateways and repeaters.

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
A gateway is required to run its radio 24 hours a day, to receive any signal that needs to
be uploaded to an IoT Cloud. This will typically require a current of about 300mA at 12
volts. The current research and implementation phase of this infrastructure requires a
typically large 200 watt PV panel, with associated heavy 12 volt deep cycle lead-acid
battery.
Figure 8 Example output from one node

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Figure 9 Example of a SmartTrap Node Circuit Board
Repeaters are used to pick up the signal from a node then transmit it to a gateway within
range, or another repeater.
Econode is developing an ‘IoT Power Board’. This will allow maximum efficiency,
flexibility, reliability and monitoring capability to support improved power management in
future designs. One of these boards is shown in Figure 10 Example of an Econode
Power Board. This is a circuit board that will be mounted in the same case as a sensor
node and will manage and monitor power from a variety of sources in a very efficient
package.
Econode intends to make this available as an open source building block to other
makers both commercial and non-commercial or maker space.
There is a significant amount of complexity with the IoT power board most of which is
about making the device safe and robust.
• Each battery is interdependently monitored for thermal run away.
• Over voltage / Over current / Under voltage protection is included.
• Reverse battery or solar panel polarity will be detected.
• Over / Under temperature charge will lead to cut out.
• Moisture ingress will be monitored and notified.
• Power path options will be monitored as optimized as configured.

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Figure 10 Example of an Econode Power Board
Action on the currently identified gateway, repeater and secondary power problem is
focused on building a ‘mini-gateway’. This device will use 3.3 volts, continuously
drawing about 20 mA while listening for signals from nodes. When repeating a signal to
a gateway or another mini-gateway it will use about 70 mA. When it is required to use a
4G cellular network to transmit data (in the absence of a gateway device), it will use
about 120 mA for a short period of time.
These power requirements will fall between those of a node and a gateway. The mini-
gateway will require rechargeable batteries and solar panels, however Econode is
confident that it will weigh under 1.5 Kg, including batteries, solar panel, case, mounting
brackets and any other required parts. This allows for relatively easy field deployment
compared the requirements for a conventional gateway described earlier in this section.
6. Further Research & Conclusion
This paper has described the operating context in which most outdoor IoT sensors
operate in isolated parts of New Zealand. The authors then described their own
experiences, using an Iterative Action Research model. This final section will suggest
some additional areas of study and conclude with a view on the place of outdoor IoT
sensors in the future.
6.1 Further Research
This paper has used literature from many sources, each approaching sensor network
power requirements in a different way in order to address questions relevant to their
specific requirements. Few peer reviewed publications come from New Zealand despite
there being a clear need and enthusiastic support for more use of sensors.
The authors believe that the academic community can contribute more to New Zealand
conservation values and renewed biodiversity, particularly through Predator Free 2050,
by teaming with industry and non-commercial participants in this field. More activity

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
mapped on to a standard research method, and published in open-access sources, will
result in improved outcomes across for this national campaign.
6.2 Conclusion
The authors have outlined one part of the broad range of work we have undertaken in
support of the Predator Free campaign. We hope that the work will be useful to others
working in this field and that making our work accessible in this way will encourage
those in the field to follow suit and also encourage younger members of our
communities to consider taking up science and engineering as an interesting and
worthwhile career.
7. References
Ayala-Ruiz, D., Castillo Atoche, A., Ruiz-Ibarra, E., Osorio de la Rosa, E., & Vázquez
Castillo, J. (2019). A Self-Powered PMFC-Based Wireless Sensor Node for Smart
City Applications. Wireless Communications and Mobile Computing, 2019,
e8986302. https://doi.org/10.1155/2019/8986302
Belhadj-Yahya, C. (2010). Performance monitoring of solar stand alone power systems.
2010 IEEE International Energy Conference, 412–416.
https://doi.org/10.1109/ENERGYCON.2010.5771715
Cayenne. (n.d.). Cayenne Features. Developer | MyDevices.Com. Retrieved May 29,
2021, from https://developers.mydevices.com/cayenne/features/
Craig, J., Norton, D., Moller, H., Saunders, D., & Williams, M. (2013). Enhancing our
heritage: Conservation for 21st Century New Zealanders: Ways forward from the
Tahi group of concerned scientists.
https://ourarchive.otago.ac.nz/handle/10523/5330
Curry, J., & Harris, N. (2019). Powering the Environmental Internet of Things. Sensors,
19(8), 1940. https://doi.org/10.3390/s19081940
dos Reis, G., Strange, C., Yadav, M., & Li, S. (2021). Lithium-ion battery data and where
to find it. Energy and AI, 5, 100081. https://doi.org/10.1016/j.egyai.2021.100081

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Elahi, H., Munir, K., Eugeni, M., Atek, S., & Gaudenzi, P. (2020). Energy Harvesting
towards Self-Powered IoT Devices. Energies, 13(21), 5528.
https://doi.org/10.3390/en13215528
Groenestein, C. (2021, February 25). Noisy tech project lures a win for students. Stuff -
Taranaki Daily News. https://www.stuff.co.nz/taranaki-daily-
news/news/300236205/noisy-tech-project-lures-a-win-for-students
Hipkins, R. (2009). Engaging students in science | New Zealand Council for Educational
Research. https://www.nzcer.org.nz/research/publications/engaging-students-
science
IEC. (2016). Internet of Things: Wireless Sensor Networks.
https://www.iec.ch/basecamp/internet-things-wireless-sensor-networks
Jaycar. (n.d.). Eclipse AA Alkaline Batteries Bulk Pack of 40 | Jaycar Electronics New
Zealand. Retrieved May 24, 2021, from https://www.jaycar.co.nz/eclipse-aa-
alkaline-batteries-bulk-pack-of-40/p/SB2332
Kim, W., Ling, T.-W., Lee, Y.-J., & Park, S.-S. (Eds.). (2001). The Human Society and
the Internet: Internet Related Socio-Economic Issues: First International
Conference, Human.Society.Internet 2001, Seoul, Korea, July 4-6 2001.
Proceedings. Springer-Verlag. https://doi.org/10.1007/3-540-47749-7
Kock, N. F., McQueen, R. J., & John, L. S. (1997). Can action research be made more
rigorous in a positivist sense? The contribution of an iterative approach. Journal
of Systems and Information Technology, 1(1), 1–23.
https://doi.org/10.1108/13287269780000732

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Lerner, J., & Tirole, J. (2005). The Economics of Technology Sharing: Open Source and
Beyond. Journal of Economic Perspectives, 19(2), 99–120.
https://doi.org/10.1257/0895330054048678
Mattern, T., & Seed, R. (2020). NZ Penguin Initiative, Q4 2020 Report, Thomas Mattern
& Richard Seed. 22.
Merriam Webster. (n.d.). Definition of BATTERY. Retrieved May 24, 2021, from
https://www.merriam-webster.com/dictionary/battery
Okkeh, Y., Masoud, B., & Batarseh, M. G. (2018). Addressing the Design Stages of an
Underground Energy Harvesting System from the Unconventional Source of
Street Moving Vehicles. 2018 Energy and Sustainability for Small Developing
Economies (ES2DE), 1–6. https://doi.org/10.1109/ES2DE.2018.8494297
PF2050. (2020). Predator Free 2050 Annual Report.
https://pf2050.co.nz/app/uploads/2020/10/PF2050-Limited-Annual-Report-
2020.pdf
Placke, T., Kloepsch, R., Dühnen, S., & Winter, M. (2017). Lithium ion, lithium metal,
and alternative rechargeable battery technologies: The odyssey for high energy
density. Journal of Solid State Electrochemistry, 21(7), 1939–1964.
https://doi.org/10.1007/s10008-017-3610-7
Revolution. (2021). PICAXE-08M2 microcontroller (AXE007M2)—Hardware—PICAXE.
https://picaxe.com/hardware/picaxe-chips/picaxe-08m2-microcontroller/
Straka, T. M., Fritze, M., & Voigt, C. C. (2020). The human dimensions of a green–
green-dilemma: Lessons learned from the wind energy—Wildlife conflict in
Germany. Energy Reports, 6, 1768–1777.
https://doi.org/10.1016/j.egyr.2020.06.028

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Susman, G. I., & Evered, R. D. (1978). An Assessment of the Scientific Merits of Action
Research. Administrative Science Quarterly, 23(4), 582.
https://doi.org/10.2307/2392581
Taranaki Mounga. (2020, July 5). South Taranaki students sound off on stoats.
https://www.youtube.com/watch?v=dqfi-GTrOPE
TEDx Talks. (2016, December 15). Using Moore’s Law to bring back New Zealand’s
native birds | Grant Ryan | TEDxChristchurch.
https://www.youtube.com/watch?v=n9PKUMHDbpw&t=170s
Towns, D. R., Daugherty, C. H., & Cree, A. (2001). Raising the prospects for a forgotten
fauna: A review of 10 years of conservation effort for New Zealand reptiles.
Biological Conservation, 99(1), 3–16. https://doi.org/10.1016/S0006-
3207(00)00184-1
Transforma Insights. (2020). IoT connected devices by vertical 2030. Statista.
https://www.statista.com/statistics/1194682/iot-connected-devices-vertically/
Wanaka Sun. (2017, May 25). Wanaka Sun | 25—31 May 2017 | Edition 819. Wanaka
Sun, 7.
Wang, H., Pan, C., Wang, Y., Xia, H., & Yu, L.-D. (2019). Comparison of Interface
Circuits for Piezoelectric Wind Energy Harvesting from Galloping Oscillation.
2019 13th Symposium on Piezoelectrcity, Acoustic Waves and Device
Applications (SPAWDA), 1–5. https://doi.org/10.1109/SPAWDA.2019.8681793
Wang, L., Todaria, P., Pandey, A., O’Connor, J., Chernow, B., & Zuo, L. (2016). An
Electromagnetic Speed Bump Energy Harvester and Its Interactions With
Vehicles. IEEE/ASME Transactions on Mechatronics, 21(4), 1985–1994.
https://doi.org/10.1109/TMECH.2016.2546179

This quality assured paper appeared at the 12th Annual Conference of Computing and Information Technology Research and Education New Zealand
(CITRENZ 2021) & 34th Annual Conference of the National Advisory Committee on Computing Qualifications, Wellington, July 14-16, alongside ITx 2021.
Watanabe, T. (2001). Merits of Open-Source Resolution to Resolve a Digital Divide in
Information Technology. Proceedings of the First International Conference on
The Human Society and the Internet - Internet Related Socio-Economic Issues,
92–99.
Way, M., & Way, A. (n.d.). About Econode. Retrieved May 29, 2021, from
https://www.econode.nz/aboutus
Whittingham, M. S. (2012). History, Evolution, and Future Status of Energy Storage.
Proceedings of the IEEE, 100(Special Centennial Issue), 1518–1534.
https://doi.org/10.1109/JPROC.2012.2190170
ZIP Report to 30 June 2015 (p. 17). (2015). http://zip.org.nz/s/ZIP-Report-to-30-June-
2015.pdf