sensors

Article
Development of a Low-Power Underwater NFC-Enabled Sensor
Device for Seaweed Monitoring
Caroline Peres *, Masoud Emam, Hamed Jafarzadeh , Marco Belcastro and Brendan O’Flynn

Tyndall National Institute, University College Cork, T12 R5CP Cork, Ireland; [email protected] (M.E.);
[email protected] (H.J.); [email protected] (M.B.); [email protected] (B.O.)
* Correspondence: [email protected]

Abstract: Aquaculture farming faces challenges to increase production while maintaining welfare
of livestock, efficiently use of resources, and being environmentally sustainable. To help overcome
these challenges, remote and real-time monitoring of the environmental and biological conditions of
the aquaculture site is highly important. Multiple remote monitoring solutions for investigating the
growth of seaweed are available, but no integrated solution that monitors different biotic and abiotic
factors exists. A new integrated multi-sensing system would reduce the cost and time required
to deploy the system and provide useful information on the dynamic forces affecting the plants
and the associated biomass of the harvest. In this work, we present the development of a novel
miniature low-power NFC-enabled data acquisition system to monitor seaweed growth parameters





 in an aquaculture context. It logs temperature, light intensity, depth, and motion, and these data can
be transmitted or downloaded to enable informed decision making for the seaweed farmers. The
Citation: Peres, C.; Emam, M.;
device is fully customisable and designed to be attached to seaweed or associated mooring lines. The
Jafarzadeh, H.; Belcastro, M.; O’Flynn,
B. Development of a Low-Power
developed system was characterised in laboratory settings to validate and calibrate the embedded
Underwater NFC-Enabled Sensor sensors. It performs comparably to commercial environmental sensors, enabling the use of the device
Device for Seaweed Monitoring. to be deployed in commercial and research settings.
Sensors 2021, 21, 4649. https://
doi.org/10.3390/s21144649 Keywords: aquaculture; sensor system; data logger; seaweed monitoring; inertial motion unit

Academic Editors: Elfed Lewis,

Cian O’Mathuna, John Barton,
Gerald Farrell, Joan Condell, 1. Introduction
Thomas Newe, Alison Keogh and
In aquaculture, maintaining the health and welfare of livestock, optimising resources,
Ciprian Briciu-Burghina
and sustainable growth are the current challenges in the continued efforts to balance eco-
nomics and environmental sustainability in the farming practices of the future. One promis-
Received: 31 May 2021
Accepted: 3 July 2021
ing solution is to adopt the practice of integrated multi-trophic aquaculture (IMTA) [1,2].
Published: 7 July 2021
IMTA cultivates different marine species in the same site, taking advantage of using by-
products (such as waste and uneaten food) from one species as inputs (fertilizer, food,
Publisher’s Note: MDPI stays neutral
and energy) for the growth of other species. This is more sustainable than monoculture
with regard to jurisdictional claims in
aquaculture sites, due to its lower environmental impact, product diversification, spatial
published maps and institutional affil- optimisation, and better management practices [1,2]. However, to optimise the production
iations. and management system, and to understand how the species interact with each other
and with the environment, technology to remotely monitor environmental and biological
conditions is needed.
A review of optical remote monitoring techniques for kelp is described in
Copyright: © 2021 by the authors.
Shroeder et al. [3] where the authors describe methods that are useful for determining
Licensee MDPI, Basel, Switzerland.
biomass and temporal trends of kelp communities. These techniques involve image ac-
This article is an open access article
quisition of an area to detect floating macroalgae, and different species have different
distributed under the terms and spectral responses to image sensors due to their different morphology and colour. The
conditions of the Creative Commons authors highlight the need for ground-truthing the data and the need for image processing
Attribution (CC BY) license (https:// algorithms that can be computationally expensive. Another review of remote monitoring
creativecommons.org/licenses/by/ methods is described in Bennion, Fischer, and Brodie [4], and it includes other techniques
4.0/). such as monitoring via Lidar and Sonar.

Sensors 2021, 21, 4649. https://doi.org/10.3390/s21144649 https://www.mdpi.com/journal/sensors

Sensors 2021, 21, 4649 2 of 19

The sensing techniques described in these papers are useful to monitor large areas
for macro-algae distribution and to indirectly derive the biomass in the area. However,
the spatial resolution of these techniques is limited, as an aquaculture site can be covered
by just a few pixels in the image. Ottinger, Clauss, and Kuenzer [5] argue that very high
resolution sensors are better suited for aquaculture site mapping and monitoring.
Another potential issue with these optical remote monitoring techniques is the over- or
underestimation of the monitored variables. For example, Meneghesso et al. [6] identified a
mismatch between the remote monitoring and deployed in situ temperature loggers. Other
studies [7,8] found the same problem. This highlights the need for an in situ deployed
sensor to accurately monitor variables such as temperature.
Different abiotic and biotic factors affect the aquaculture farm production. In particu-
lar, wave exposure and water motion affect the growth of seaweed, but the effect is not
completely understood [9–12]. Since most studies of this effect use non-direct measure-
ments of wave exposure, such as wave exposure index [13] derived by wind speed and
incidence, they do not provide good correlation with actual wave measurements [14].
The accurate measurement of these environmental parameters is needed to design and
validate ecosystem models for IMTA production, such as Fan et al. [15], due to the current
challenge of IMTA in understanding how the species interact with each other and the
environment. To validate these models and improve production efficiency, environmental
sensors—such as temperature, light, water quality (such as the concentration of dissolved
oxygen, water pH, salinity, turbidity, concentration of pollutants), nutrient (dissolved
nitrogen) availability, and water motion—are needed [16,17].
Monitoring of water quality and some environmental parameters—such as
temperature—in an aquaculture setting is well supplied by commercial off-the-shelf sen-
sors. Their measurement parameters include temperature, light radiation, and water
characteristics (dissolved oxygen, pH, salinity, nitrogen concentration). However, these
sensors are expensive and usually not co-located with the seaweed growth site. Addition-
ally, these sensors are usually big, require cables and a power supply, and measure only
one parameter, which increases the power consumption, maintenance, and cost [18,19]. For
example, Visch, Nylund, and Pavia [10] used Onset HOBO Pendant UA-002-64 loggers [20]
to monitor and log temperature. However, this commercial sensor is big, only measures
temperature and light, and needs to be wired which limits the deployment in areas where
the accessibility is hard or there is no power.
In the case of measuring of wave exposure and water motion rates, sensors such as
acoustic doppler current profilers (ADCP) or water buoys are generally used. However,
these sensors are usually expensive and do not allow for multiple point deployments.
Some cheaper solutions were developed by [14,19,21,22] using accelerometers, but none of
them provide a complete multi-sensor integrated solution.
Judge, Choi, and Helmut [23] review the current logger technology for intertidal
environments. Although not the same application, the environment is similar to the one
seaweed is cultivated, and the loggers used to monitor are the same. They argue that
the current loggers available are limited in: (1) construction—as the attachment requires
large amounts of epoxy that interfere with temperature readings; (2) lack of real-time data
due to the wireless communications constraints of underwater environments and hard to
download data; (3) miniaturization. The dataloggers used in the studies compiled by Judge,
Choi, and Helmut [23] are: (a) iButton devices by Maxim Integrated [24] that, although
developed for harsh marine environment, require a cable to transmit data to a receptor; (b)
HOBO TidbiT by Onset [25] that is small and waterproof. Both solutions, however, only
log temperature.
Other sensors to monitor aquaculture species were developed using easy-to-use
platforms such as Arduino [26,27]. The device developed by Beddows and Mallon [27] is
able to log only temperature with an operational lifetime of up to 1 year in non-rechargeable
batteries. The device is housed in a big enclosure that is able to withstand the harsh marine
environment. The data collected need to be downloaded by recovering the device.
Sensors 2021, 21, 4649 3 of 19

For wave measurement, Knight et al. [26] developed an IoT-enabled tide gauge with a
pressure sensor and an Arduino. This sensor is designed to be placed in a fixed position
and requires cables to a power source. Kennedy et al. [22] developed a miniature wave
measurement device as a replacement for wave buoys. It has a 6-degrees-of-freedom (DoF)
inertial measurement unit (IMU) and a Zigbee wireless connection.
All the solutions discussed above have some problems in common, with varying
degrees: big size that limits the deployment method; power requirements such as cables,
non-rechargeable batteries, external power supplies; data recovery, as most do not transmit
data wirelessly (requiring cable loggers that store data for transmission) or limit deploy-
ment in hard-to-reach locations, as device recovery is challenging; limited to one sensor
type per enclosure—as far as we know, no miniaturised multi-sensor modality solution is
available for seaweed monitoring.
In this work, we present a novel miniature low-power near-field communication
(NFC)-enabled data acquisition system to monitor aquaculture species. This sensor system
monitors temperature, light intensity, depth, and motion, logs the data internally, and can
transmit the logged data via NFC (to a smartphone, for example). It also has an internal
machine-learning-enabled microcontroller, which can be used to analyse data internally.
The mechanical enclosure design allows it to be attached to IMTA species for data
acquisition. The device is designed to be attachable to seaweed and kelp blades or stipes: it
has a texture on the bottom side for gluing onto the blades; it also has holes for threading
safety threads to secure the device to the mooring line or to tie it to the stipe. The tem-
perature and depth sensor has a direct interface with water, which allows the sensor to
accurately measure water temperature even with glue or epoxy attachments.
The sensor device can communicate with NFC-enabled readers (such as smartphones)
to configure the sensors with custom sampling frequencies, communicate status, and to
download data.
An IP-rated enclosure for the multi-modal sensing system was developed, and the
device was characterised in lab experiments to verify if it complies with the requirements
outlined above and to determine its battery lifetime with different sensor sampling condi-
tions prior to deployment.
This paper is organised as follows: In Section 2, we present the design rationale and
architecture of the device. We explain in detail the hardware, enclosure, antenna, firmware,
and smartphone application developed. In Section 3, we present the methods and results
for the characterisation of the sensors embedded in the device, and the power consumption
measurements. In Section 4, we discuss the results and the relevancy of the sensor device
developed for the aquaculture industry.

2. Materials and Methods

2.1. Hardware Architecture
In conjunction with stake holders and end users (seaweed farm owners) and knowl-
edge from current state-of-the-art, the requirements for the device were defined as follows:
• The device must be as small as possible as to not disturb the seaweed physiology; be
reusable to reduce environmental waste and pollution; waterproof and resistant to the
marine environment.
• The sensors should be configurable regarding sampling frequency and other relevant
parameters to allow greater customisation in the deployment. The device should be
able to log temperature, depth, movement via acceleration, and light intensity on a
single small package.
• The device must have a battery and wireless communication to be able to collect the
logged data remotely.
• The battery should be rechargeable to reduce waste, and its lifetime should be maximised
to allow longer periods of deployment without the need for farm operator interaction.
Based on the user requirements above, the device developed is composed of various
sensor modalities, a microcontroller as a supervisor for the device, an external flash memory
 • The device must have a battery and wireless communication to be able to collect the
logged data remotely.
• The battery should be rechargeable to reduce waste, and its lifetime should be max-
Sensors 2021, 21, 4649 imised to allow longer periods of deployment without the need for farm operator
4 of 19
interaction.
Based on the user requirements above, the device developed is composed of various
sensor modalities, a microcontroller as a supervisor for the device, an external flash
to save collected data, and a radio-frequency identification (RFID)/NFC front-end to
memory to save collected data, and a radio-frequency identification (RFID)/NFC front-
communicate with NFC readers while deployed. The system block diagram for the device
end to communicate with NFC readers while deployed. The system block diagram for the
is shown in Figure 1.
device is shown in Figure 1.

Figure1.1.System
Figure Systemblock
blockdiagram
diagramfor
forthe
thedevice
devicehardware.
hardware.

When selecting
When selecting the the components
componentsfor thethe
for system,
system,the the
mainmain
constraints were were
constraints size, power
size,
power consumption, and compatibility with the other components and the systemvoltage
consumption, and compatibility with the other components and the system main main
value. value.
voltage
Duetotothe
Due thesize
sizeconstraints
constraintsidentified
identifiedand andthe theextended
extendedoperational
operationallifetime
lifetimerequire-
require-
ment,the
ment, the battery
battery technology
technology chosen chosenwas waslithium-ion
lithium-ionpolymer polymer (LiPo).
(LiPo).TheTheadditional
additionalben-
efit of this
benefit typetype
of this of battery
of battery is theissmall weight
the small and the
weight and rechargeability. The nominal
the rechargeability. The nominalvoltage
of a LiPo
voltage of abattery cell is around
LiPo battery 3.7 V, so
cell is around 3.7itV,can
so it becan
usedbe to power
used integrated
to power circuits
integrated in the
circuits
in3.0–3.3 V range.
the 3.0–3.3 We chose
V range. We 3.0 choseV as3.0theVmain
as the voltage of the circuit
main voltage of thetocircuit
extendtothe power-on
extend the
time, letting
power-on time,theletting
battery thedischarge as much as
battery discharge as much
possible. The battery
as possible. Thechosen
batteryhaschosena manage-
has a
ment systemsystem
management embedded embedded that protects the battery
that protects fromfrom
the battery over-discharge,
over-discharge, overcharge,
overcharge, and
and short
short circuits.
circuits. It isItais300
a 300
mAh mAh capacity
capacity LiPoLiPo battery
battery withwith dimensions
dimensions of of approximately
approximately 36
36mmmm × 12
× 12 mm mm × 6.5
× 6.5 mm. mm.
AAbattery
batterycharger
charger ICICwas wasadded
added to the design
to the to control
design the charging
to control of theof
the charging battery over
the battery
the USB connector. The IC used for this is has a small footprint,
over the USB connector. The IC used for this is has a small footprint, allows the configu-allows the configuration of
the charging
ration of thecurrent
charging viacurrent
an external
via anresistor
external connected to one of its
resistor connected topins,
one ofand itsitpins,
has aand
pinit
input
has afor
pinainput
thermistor for thermalfor
for a thermistor shutdown in case of high
thermal shutdown temperature
in case sensed. sensed.
of high temperature
The
Thebattery
batteryisisalways
alwaysconnected
connectedand andpowers
powersthe thewhole
wholesystem
systemvia viaa alow-dropout
low-dropout
regulator (LDO) when in normal and sleep modes. To conserve
regulator (LDO) when in normal and sleep modes. To conserve battery energy when battery energy when not not
in
the data collection mode, the device has a shutdown mode that
in the data collection mode, the device has a shutdown mode that supplies power only to supplies power only to the
wake-up
the wake-upsubsystem.
subsystem.
An
An ON/OFF controller
ON/OFF controller[28] [28]waswasadded
addedto tothethedesign
designtotomanage
managethe theshutdown
shutdownmode mode
bybyturning
turningon onandandoff offthethemainmainsystem
systemLDO.LDO.The Thecontroller
controllerturns
turnson onthe
themain
mainsystem
system
power
powerregulator
regulatorvia viatwotwoevents:
events:(1) (1)aawake-up
wake-upsignal signalfrom
fromthetheNFCNFCapplication
applicationoror(2) (2)byby
connecting the device to a
connecting the device to a USB host. USB host.
The
Thesensors
sensorschosen
chosenfor forthe device
the device were
were carefully
carefullyselected to be
selected to as
be small form
as small factor
form as
factor
possible, with very low power consumption and compatibility
as possible, with very low power consumption and compatibility with the system voltage with the system voltage of
3.0 V. Another factor that went into the sensor selection is the configurability and resolution.
The movement sensor used in the system integrates a 3D accelerometer and a 3D
gyroscope in a very small package with very low power consumption and low noise. The
IC also has an embedded machine learning core that can be used to filter and detect features
in the motion data, reducing the processing needed in the controller.
For temperature and pressure sensing, the device chosen has a very small form factor
and is optimised for water depth measurements. It has a high depth resolution of 0.2 cm
Sensors 2021, 21, 4649 5 of 19

in water, operating pressure range of 3 MPa, and low power consumption. The pressure
sensing opening is built for chemical endurance in hash liquid media (such as seawater),
and it can be watertight using an O-ring. In addition to pressure sensing, it provides
temperature sensing as well ranging from −20 ◦ C to 85 ◦ C and resolution of 0.0022 ◦ C.
For light/lux sensing, a colour + clear light sensor with embedded IR (infrared)
blocking filter was selected. A colour sensor was preferred due to the possibility of
correlating the RGB channel values with the photosynthetically active radiation (PAR)
sensors, in addition to the clear (white light) channel that correlates with the lux values.
The sensor chosen has programmable integration time and gain, which allows it to measure
values in different light intensity environments. Other key factors are the very small size
and the low power consumption.
A microcontroller (MCU) was added to the design to control and manage the whole
system, transfer data from sensors to external memory, and communicate with hosts. The
requirements for the MCU include low power consumption, small size, high processing
power, large RAM, enough peripherals for all the sensors and external circuits, and floating
point. The MCU chosen was the STM32L4R5QII6 [29], due to its low power, machine
learning compatibility, and high CoreMark and ULPMark scores [30,31] compared to other
MCUs of same power consumption.
To make sure there is enough space to save all sensor data collected, the SPI Flash
memory with the largest memory size possible was selected. For a serial SPI Flash, the
largest memory capacity available in a commercial off-the-shelf IC is 1 GB. This is roughly
equivalent to 2 weeks of continuous accelerometer and gyroscope data (with sampling
frequency of 52 Hz).
To communicate download data collected and wake-up the system, two communica-
tion interfaces are present in the device design: (1) a wired interface using a USB type C
connector and (2) a wireless communication interface using NFC/RFID at 13.56 MHz.
The USB connection can be used to communicate with the MCU, download data,
configure the system, update the firmware, and wake-up the system. USB was chosen
because of its ease of use and compatibility with any personal computer and laptops. The
USB connector selected is waterproof with an O-ring around its external chassis, which
adds another layer of protection to the device when using it in wet areas. The USB is
also used to provide the power to the device, powering the main system voltage and/or
charging the battery.
Water, especially seawater, is a challenging environment for wireless communica-
tion [32]. Radio-frequency (RF) communications suffer from propagation loss due to the
water salinity, and this effect is directly correlated with the frequency of transmission as
the higher salt concentration makes the water conductive in these high frequencies [33,34].
On the other hand, lower frequencies—which would be less affected by the attenuation—
require large antennas [34]. To mitigate this problem, one can use electromagnetic fields in
the near-field (the transmission principle behind some RFID tags and near-field communi-
cation (NFC) devices), also called magnetic induction (MI) [35,36]. These systems transmit
using the magnetic field, which suffers less attenuation in seawater than the electric field
used in other RF communications [35]. The lower frequency of these systems also lowers
the attenuation factor [36].
Near-field communication (NFC) is a standard of communication on the high-frequency
(HF) RFID at 13.56 MHz. In an RFID system, the transmitter (also called reader) generates
an electromagnetic field from its loop antenna that induces a voltage in the receiver (or
tag) antenna located inside this field. The reader-tag pair then works like two coupled
inductors, and the impedances at each side are reflected back to the opposite side, which is
how data are transmitted [37]. An added benefit of this system is that the induced voltage
in the antenna can be used to power up the tag, effectively harvesting energy from the field.
The NFC standard is widely used today, and a great number of modern smartphones
have an NFC reader/writer controller embedded to be used as a smart wallet. This can
then be leveraged to be an easily accessible and operable reader for the device through
 ceiver (or tag) antenna located inside this field. The reader-tag pair then works like two
coupled inductors, and the impedances at each side are reflected back to the opposite side,
which is how data are transmitted [37]. An added benefit of this system is that the induced
voltage in the antenna can be used to power up the tag, effectively harvesting energy from
the field.
Sensors 2021, 21, 4649 6 of 19
The NFC standard is widely used today, and a great number of modern smartphones
have an NFC reader/writer controller embedded to be used as a smart wallet. This can
then be leveraged to be an easily accessible and operable reader for the device through
thedevelopment
the developmentof ofsuitable
suitablesmart
smartphone
phoneapps.apps.Another
Anotherbenefit
benefitisisthat
thatthetheNFC
NFCreader
reader
providesmost
provides mostofofthe thepower
powerusedusedininthethecommunication,
communication,asasstated statedbefore,
before,which
whichgreatly
greatly
reduces
reducesthethepower
powerconsumption
consumptionfor forthis
thiscommunication
communicationinterface.
interface.On Onthetheother
otherhand,
hand,the the
range
rangeofofNFC
NFCisislimited
limitedby bythe
thereader
readertransmitted
transmittedpower powerand andantenna
antennasizes
sizes[36,38].
[36,38].TheThe
NFC
NFCstandard
standard[39] [39]limits
limitsthe
thestrength
strengthof ofthe
themagnetic
magneticfieldfieldcreated
createdby byaareader,
reader,effectively
effectively
limiting
limitingthe
therange
rangefor forcommercial
commercialsystems
systemstotoaround
around2020cm. cm.
The
Theinterface
interfacefor forthe
theNFC
NFCchosenchosenwas wasthetheST25DV
ST25DVchip chiptransponder
transponder[40], [40],which
whichon on
top
topofofhaving
havingan aninternal
internalEEPROM,
EEPROM,can canalso
alsocommunicate
communicatevia viaaamailbox
mailboxsystem
systemthat thatcancan
transmit
transmitup uptoto256
256bytes
bytesper
permessage
messageon ondemand.
demand.This Thismakes
makesthe thecommunication
communicationbetween between
reader and transponder more efficient, as there is no need to
reader and transponder more efficient, as there is no need to read/write the internal read/write the internal
EEPROM.
EEPROM.The TheST25DV
ST25DVchip chipis is
connected
connected to to
thethe
MCU MCU viavia
I2CI2Cserial interface,
serial allowing
interface, allowingit toit
act as a bridge between the MCU and the
to act as a bridge between the MCU and the NFC reader. NFC reader.
This
Thischip
chipcancanalsoalsocommunicate
communicate with
withthethe
NFC reader
NFC battery-less.
reader ThisThis
battery-less. means that the
means that
system can be
the system caninbeshutdown
in shutdownmodemode and still
andbe able
still be to respond
able to respondto commands.
to commands. We use Wethisuse
feature to create a wake-up circuit that the NFC chip turns on when
this feature to create a wake-up circuit that the NFC chip turns on when receiving a com- receiving a command
from
mand thefrom
reader.
the In this mode,
reader. In thisthemode,
battery theonly powers
battery onlythepowers
real-timetheclock in theclock
real-time MCUinand the
the wake-up circuit, effectively reducing energy consumption when
MCU and the wake-up circuit, effectively reducing energy consumption when not in use. not in use.
The
Theelectronics
electronicsare aremounted
mountedon onvery
verysmall
smallcustom-fabricated
custom-fabricatedprinted printedcircuit
circuitboards
boards
(PCB)
(PCB) with 4 layers and impedance matching for the USB line. Figure 2 showsthe
with 4 layers and impedance matching for the USB line. Figure 2 shows thefinal
final
PCB design.
PCB design.

Figure2.2.3D
Figure 3Dview
viewofofthe
thefinal
finaldesigned
designedPCB.
PCB.The
Theorange
orangeblock
blockrepresents
representsthe
thebattery.
battery.

2.2.RFID/NFC
2.2. RFID/NFCAntenna
AntennaDesign
Design
We
Wedesigned
designedaacustom
customantenna
antennaforforthis
thisapplication,
application,asasno
noCOTS
COTSantenna
antennahad hadthethesize
size
required
requiredfor
forthe device.
the device.The antenna
The antenna was designed
was to be
designed toabe
flexible printed
a flexible circuit
printed boardboard
circuit that
connects to theto
that connects main
the PCB
mainviaPCBa flexible printedprinted
via a flexible circuit (FPC)
circuitconnector. Its size (excluding
(FPC) connector. Its size (ex-
the connection
cluding cable) is 40
the connection × 13ismm
cable) 40 ×and was and
13 mm chosen
wasdue to the
chosen overall
due to thePCB andPCB
overall battery
and
size. The antenna parameters were designed using ST Antenna eDesign Suite
battery size. The antenna parameters were designed using ST Antenna eDesign Suite [41] [41] to match
the expected
to match impedance
the expected for the NFC
impedance chip
for the NFCused
chipinused
the design (ST25DV).
in the design To design
(ST25DV). the
To design
antenna and calculate
the antenna its parameters,
and calculate the procedure
its parameters, in STMicroelectronics
the procedure application
in STMicroelectronics note
application
Sensors 2021, 21, x FOR PEER REVIEW
AN2866 [42] was followed. Figure 3Figure
shows3the pattern 7 of 20
note AN2866 [42] was followed. shows the design
patternfor the rectangular
design antenna
for the rectangular
flexible
antennaPCB.
flexible PCB.

Figure3.3.Flexible
Figure FlexiblePCB
PCBtrack
trackdesign
designfor
forthe
theantenna
antennaresonant
resonanttoto13.56
13.56MHz
MHzwhen
when combined
combined with
with the
the ST25DV NFC chip (L = 4.88 µH).
ST25DV NFC chip (L = 4.88 µH).

The internal capacitance of the NFC chip used (ST25DVxxx) for the antenna pins is
28.5 pF. To make a resonant antenna at the 13.56 MHz frequency, we can use the following
relationship to determine the inductance of the antenna:
 Figure 3. Flexible PCB track design for the antenna resonant to 13.56 MHz when combined with
Sensors 2021, 21, 4649 the ST25DV NFC chip (L = 4.88 µH). 7 of 19

The internal capacitance of the NFC chip used (ST25DVxxx) for the antenna pins is
28.5 pF.
TheTo make acapacitance
internal resonant antenna
of the NFCat the
chip13.56
usedMHz frequency,
(ST25DVxxx) for we
the can use the
antenna pinsfollowing
is
relationship to determine
28.5 pF. To make a resonant the inductance
antenna of the
at the 13.56 MHz antenna:
frequency, we can use the following
relationship to determine the inductance of the antenna:
1
𝑋 2πf𝐿 0
1 2πf𝐶
X = 2πfL + =0
From the above equation, we calculate2πfC
the required antenna inductance to be L = 4.83
µH. Using
From thetheabove
ST Antenna
equation,eDesign Suite,
we calculate thethe parameters
required antennafor conductor
inductance to bewidth and
L = 4.83 µH.spacing
were
Usingcalculated for theeDesign
the ST Antenna antennaSuite,
size, the
butparameters
due to the for
small size, a compromise
conductor width and spacingwas needed
were
on thecalculated
inductance for the antenna
value. The size,
real but due to thecalculated
inductance small size, awas
compromise
L = 4.88 wasµH.needed
To solve the
on the inductance
discrepancy value. The
in inductance real inductance
value and possible calculated
fabrication L = 4.88 µH. the
was deviations, To solve
mainthePCB was
discrepancy in inductance value and possible fabrication deviations,
designed with space to solder a capacitor in parallel with the antenna connector the main PCB wasto com-
designed with space to solder a capacitor in parallel with the antenna connector to complete
plete the resonant circuit.
the resonant circuit.
Once
Once fabricated, theantenna
fabricated, the antenna inductance
inductance waswas measured
measured usingusing
a VNA, a VNA,
and theand the result
result
can
canbebe seen in Figure
seen in Figure4.4.Converting
Converting thethe impedance,
impedance, we find
we find thatinductance
that the the inductance
of the of the
antenna is indeed 4.88
antenna is indeed 4.88 µH. µH.

Figure
Figure4.
4. Smith chartand
Smith chart andmeasured
measured impedance
impedance forfabricated
for the the fabricated
flexibleflexible PCB antenna
PCB antenna at 13.56
at 13.56 MHz.
MHz.
2.3. Enclosure Design
The enclosure
2.3. Enclosure Designwas designed using the PCB 3D model and the battery size as base
sizes. A screw-on cap is used to access the USB-C connector that is also waterproof.
The The
backenclosure
is designedwas
withdesigned using
relief details the the
to aid PCB 3D model
attachment to and the battery size
the seaweed/kelp leafas base
sizes. A screw-on cap is used to access the USB-C connector that is also
using glue. The whole enclosure is waterproofed by using gaskets and screws, as well as waterproof. The
back
usingisadesigned withresin
polyurethane relief details
(PUR) to aid
inside the attachment
the device to the
to protect the seaweed/kelp
electronics leaf using
from water
glue. TheThe
ingress. whole
holesenclosure is waterproofed
on the sides provide a spaceby to using
thread gaskets and screws,
a line securing as well
the device as using
to the
mooring lines in the farm. This is needed to make sure the device is not lost and becomes
an environment pollutant unintentionally. The 3D design of the enclosure can be seen
in Figure 5.
 a polyurethane resin (PUR) inside the device to protect the electronics from water ingress.
The holes on the sides provide a space to thread a line securing the device to the mooring
Sensors 2021, 21, 4649 lines in the farm. This is needed to make sure the device is not lost and becomes an8 envi-
of 19
ronment pollutant unintentionally. The 3D design of the enclosure can be seen in Figure
5.

Figure5.5.3D
Figure 3Dview
viewofofthe
theenclosure
enclosuredesigned
designedfor
forthe
thedevice.
device.

2.4.
2.4.Firmware
FirmwareDesign
Design
The
Thesensor
sensordevice
devicehas three
has threemodes
modes of of
operation:
operation: shutdown,
shutdown,sleep, andand
sleep, normal. During
normal. Dur-
shutdown, all theallcomponents
ing shutdown, are turned
the components off, except
are turned for thefor
off, except NFCthechip,
NFCON/OFF controller,
chip, ON/OFF con-
and the RTC
troller, (real-time
and the clock) onclock)
RTC (real-time the microcontroller. The shutdown
on the microcontroller. mode canmode
The shutdown be entered
can be
via sending
entered viaasending
turn-offacommand to the device
turn-off command viadevice
to the eithervia
NFC or USB.
either NFC or USB.
During
Duringnormal
normaloperation,
operation,thethedevice
deviceisiseither
eithercommunicating
communicatingwith withthe
theUSB
USBhost
hostoror
the
theNFC
NFCreader,
reader,ororthethedevice
deviceisiscollecting
collectingsensor
sensordata.
data.Otherwise,
Otherwise,thethedevice
deviceenters
entersthe
the
sleep mode to save battery power while still waking up periodically to collect
sleep mode to save battery power while still waking up periodically to collect the sensor the sensor
data.
data.Figure
Figure66shows
showsthethemodes
modesofofoperation
operationandandthethetransition
transitionbetween
betweenthese
thesestates.
states.

Figure 6. State diagram for seaweed sensor operations.

Figure 6. State diagram for seaweed sensor operations.

Duringdata
During datacollection
collectionfrom
fromthe
thesensors,
sensors,thetheread
readdata
dataare
aresaved
savedininaabuffer
bufferinside
insidethe
the
MCUmemory.
MCU memory. To Toreduce
reduce power consumption, the
power consumption, the MCU
MCU only onlywrites
writesthe
thedata
datacollected
collectedin
inthetheexternal
externalflash
flashmemory
memorywhen whenthis
thisbuffer
bufferisisfull.
full.InInthis
thisway,
way,the
theexternal
externalmemory
memoryis
turned off for most of the time, and only turned on during transfers.
is turned off for most of the time, and only turned on during transfers. If the device If the device is con-
is
figured to collect data, the sensor data collection subroutine should
configured to collect data, the sensor data collection subroutine should be running evenbe running even while
it is communicating
while it is communicatingwithwith
a host. To optimize
a host. To optimizethe the
battery life,life,
battery STM32CUBEMX
STM32CUBEMX battery life
battery
estimator
life estimator hashas
been used.
been To specify
used. the timing
To specify the timingrequirements
requirements of theofestimator, the clock
the estimator, the
clock
countercounter
of theofdata
the data watchpoint
watchpoint and trace
and trace (DWT) (DWT)unit ofunit of STM32
STM32 has been
has been used.used. To
To mini-
minimize
mize the the battery
battery power
power consumption,
consumption, all non-essential
all non-essential peripherals
peripherals were
were disabled,
disabled, andand
the
the microcontroller was set to run at 16MHz to minimize the current consumption of the
CPU. In Figure 7, the STM32CubeMX battery life estimator has been shown.
 Sensors 2021, 21, x FOR PEER REVIEW 9 of 20
Sensors 2021, 21, x FOR PEER REVIEW 9 of 20

Sensors 2021, 21, 4649 9 of 19

microcontroller was set to run at 16MHz to minimize the current consumption of the CPU.
microcontroller was set to run at 16MHz to minimize the current consumption of the CPU.
In Figure 7, the STM32CubeMX battery life estimator has been shown.
In Figure 7, the STM32CubeMX battery life estimator has been shown.

Figure 7. Power consumption profile calculated by STMCubeMX battery life estimator.

Figure7.7.Power
Figure Powerconsumption
consumptionprofile
profilecalculated
calculatedby
bySTMCubeMX
STMCubeMXbattery
batterylife
lifeestimator.
estimator.
The final
The final embedded
embedded system
systemdeveloped
developedisisshown
showninin
Figure 8 with
Figure thethe
8 with individual com-
individual com-
The
ponent final
parts ofembedded
the device: system
battery,developed is shown
motherboard, and NFCin Figure 8 with
antenna, as theasindividual
well the encap- com-
ponent parts of the device: battery, motherboard, and NFC antenna, as well as the encapsu-
ponent
sulated parts
system of the
readydevice: battery, motherboard, and NFC antenna, as well as the encap-
lated system ready forfor deployment.From
deployment. Fromhereafter,
hereafter,this
thissystem
system will
will be
be called
calledAquaBit.
AquaBit.
sulated system ready for deployment. From hereafter, this system will be called AquaBit.

Figure 8. Final embedded system for seaweed monitoring before and after encapsulation for de-
ployment.
Figure 8. Final embedded system for seaweed monitoring before and after encapsulation for de-
Figure 8. Final embedded system for seaweed monitoring before and after encapsulation
ployment.
2.5. Host Application
for deployment.
To communicate with the device once it is enclosed and to download data, two host
2.5.Host
2.5. HostApplication
Application
applications were developed: (1) an Android application to be used in an NFC-enabled
ToTocommunicate
smartphonecommunicate withthe
with
and (2) a Windows thedevice
deviceonce
software once itisis
runitin enclosed
aenclosed
PC andtotodownload
and
or laptop. downloaddata,data,two
twohost
host
applications
applications wereapp
were
The Android developed:
developed: (1)an
(1)
was developed anusing
Android
Android application
application
Android totobe
Studio with be used
used
the helpinin an
ofan NFC-enabled
NFC-enabled
the Software
smartphoneand
smartphone
Development and (2)
Kit(2) aaWindows
(SDK) Windows software
software
for the ST25DV run
run
[43] ininamade
tag aPC
PCoravailable
orlaptop.
laptop.by STMicroelectronics.
The TheThe Android
Androiduses
application app
appthe was
was developed
developed
SDK using
and the using
Android Android
Android Studio
Studio with
NFC libraries with
to thehelp
the helpofofthe
communicate theSoftware
withSoftware
the
Development
ST25DV
Development Kit(SDK)
tag using
Kit (SDK)
its forfor
custom the ST25DV
thecommands
ST25DV [43]
according
[43] tag to
tag made made
its available
datasheet
available by STMicroelectronics.
[40]. It can also com-The
by STMicroelectronics.
municate
application via
The application thethe
uses standard
uses SDK NFC
theand
SDK Forum
theand theType
Android 5 standard
Android
NFC commands
NFC libraries
libraries to as
to communicate implemented
communicate by the
with
with the ST25DV
the SDK [43].
ST25DV
tag using tag using its
its custom custom commands
commands accordingaccording to its datasheet
to its datasheet [40]. It can[40].
alsoItcommunicate
can also com-
via the standard
municate NFC
via the ForumNFC
standard Type Forum
5 standard
Type commands
5 standard ascommands
implemented as by the SDK [43].
implemented by
The application
the SDK [43]. functionality is as follows: the user places the smartphone close to
the device with the application open. The application reads the device configuration and
enables the user to change it by showing the options on the screen. The user has the option
to download the logged data and upload them to the IMPAQT cloud severs, which can
Sensors 2021, 21, x FOR PEER REVIEW 10 of 20

Sensors 2021, 21, 4649 The application functionality is as follows: the user places the smartphone close to
10 of 19
the device with the application open. The application reads the device configuration and
enables the user to change it by showing the options on the screen. The user has the option
to download the logged data and upload them to the IMPAQT cloud severs, which can
be
be used
used to
to visualize
visualize the
the data
data on
on the
the IMPAQT
IMPAQT monitoring
monitoring system
system (IMS).
(IMS). The
The uploading
uploading
procedure is shown in Figure
procedure is shown in Figure 9.9.

9. Data upload procedure

Figure 9. procedure to
to the
the IMS
IMS for
for visualisation.
visualisation.

To
To communicate
communicate via via USB,
USB, aa Python
Python 3.6
3.6 application
application was
was developed
developed to to work
work with
with the
the
pyusb
pyusb package
package [44]
[44] that
that provides
provides aa wrapper
wrapper to to USB functionality in
USB functionality in Windows-32
Windows-32 environ-
environ-
ments.
ments. InIn the
the device
device side,
side, the
the USB
USB stack
stack was
was set
set to
to function
function asas aa USB
USB communications
communications
device class (CDC), using the library provided by STMicroelectronics
device class (CDC), using the library provided by STMicroelectronics for ST32L4 for ST32L4 microcon-
micro-
trollers [45].
controllers [45].
This
This Python
Python application
application (hereafter
(hereafter called
called PyHost)
PyHost) is is able
able to
to send
send commands
commands to to the
the
device, update its internal real time clock, and download data from the
device, update its internal real time clock, and download data from the device internal device internal
memory.
memory. It It also
also has
has debug
debug capabilities
capabilities and
and can
can update
update the
the firmware
firmware of of the
the device
device without
without
the access to the programmer connector.
the access to the programmer connector.
3. Results
3. Results
The final embedded system has been fully characterised in Lab settings to ensure that
The final embedded system has been fully characterised in Lab settings to ensure that
accurate measurements and datasets will be obtained when deployed in the seaweed farm
accurate measurements and datasets will be obtained when deployed in the seaweed farm
associated with the IMPAQT project [46].
associated withfunctionality
A general the IMPAQTtest project
was[46].
also performed to determine system usability. The
A general functionality test was
achieved read range for the NFC communication also performed was to determine
dependent onsystem usability.
the reader: The
different
achieved readused
smartphones range for the NFC
provided communication
different read distances. wasThedependent
maximum onachieved
the reader: different
distance on
smartphones
the air was 5 cm. used provided different read distances. The maximum achieved distance on
the air was 5 cm.
3.1. Sensors Characterisation and Calibration
3.1. Sensors
3.1.1. InertialCharacterisation
Measurement and UnitCalibration
3.1.1.To
Inertial Measurement
characterise Unit the Inertial Measurement Unit (IMU), we used a motion
and calibrate
capture (MoCap) system
To characterise and a pendulum
and calibrate the Inertial device with a gimbal
Measurement platform
Unit (IMU), attached
we used to it.
a motion
The system
capture was put
(MoCap) in movement
system by oscillating
and a pendulum the with
device pendulum
a gimbalwhile the gimbal
platform was set
attached to to
it.
aThe
specific orientation, so all 3 axes of the IMU were at least once the
system was put in movement by oscillating the pendulum while the gimbal was set focus of the oscillation.
The motion
to a specific capture (MoCap)
orientation, system
so all 3 axes of theused
IMU consists
were atofleast
10 infrared
once thecameras
focus of bytheOptitrack
oscilla-
(Prime X 13, NaturalPoint, Inc. DBA Optitrack, Corvallis, OR, USA) [47] mounted in different
tion.
positions,
The heights,
motion and distance
capture from the
(MoCap) active used
system area, pointed
consistsatofthe10same area. The
infrared position
cameras by
of the cameras
Optitrack (PrimeisX13,
shown in Figure 10.
NaturalPoint, Inc. This
DBAsystem
Optitrack,is designed
Corvallis,toOR,track
US) the position
[47] mounted of
passive or active (IR-LEDs) markers placed in the object under observation
in different positions, heights, and distance from the active area, pointed at the same area. and uses
these markersoftothe
The position determine
camerasthe positioninand
is shown orientation
Figure 10. Thisofsystem
the object in relation
is designed to to global
track the
predetermined
position of passive or active (IR-LEDs) markers placed in the object under observation can
coordinates. Each camera has a 1.3 MP resolution, and its frame rate and
be adjusted
uses betweento30
these markers and 240 FPS.
determine In this experiment,
the position and orientation passive
of thereflective
object inmarkers
relation of
to
12.5 mm diameter were used. The cameras are connected to a computer
global predetermined coordinates. Each camera has a 1.3 MP resolution, and its frame using the Motive
rate
2.2
cansoftware [48]between
be adjusted that combines
30 andthe 240data
FPS.from theexperiment,
In this cameras to recreate
passive the movement
reflective of the
markers of
object being tracked. The MoCap system was calibrated according to the instructions of
the manufacturer [49]. The calibration file was saved, as it contains the calibration error
necessary to assess the results.
 object
Sensors 2021, 21, x FOR PEER REVIEW being tracked. The MoCap system was calibrated according to the instructions 11 of of20
the manufacturer [49]. The calibration file was saved, as it contains the calibration error
necessary to assess the results.
The diameter
12.5 mm custom pendulum
were used.setup is composed
The cameras of four strings
are connected attached to
to a computer an aluminium
using the Motive
Sensors 2021, 21, 4649 rig structure via bolts and nuts and a gimbal platform with 2 DoF. Figure
2.2 software [48] that combines the data from the cameras to recreate the movement11 shows
11 of the
19
of the
detailed photo
object being of the The
tracked. gimbal platform
MoCap where
system wasthe device was
calibrated placedtowith
according the the passive re-
instructions of
flective markers attached to the corners.
the manufacturer [49]. The calibration file was saved, as it contains the calibration error
necessary to assess the results.
The custom pendulum setup is composed of four strings attached to an aluminium
rig structure via bolts and nuts and a gimbal platform with 2 DoF. Figure 11 shows the
detailed photo of the gimbal platform where the device was placed with the passive re-
flective markers attached to the corners.

Figure10.
Figure 10.Positioning
Positioning ofof the
the MoCap
MoCap system
system cameras
cameras in
in relation
relationto
tothe
thedevice.
device.Cameras
Camerasare
arehigh-
high-
lighted with red circles.
lighted with red circles.

The custom pendulum setup is composed of four strings attached to an aluminium rig
structure via bolts and nuts and a gimbal platform with 2 DoF. Figure 11 shows the detailed
photo
Figureof
10.the gimbal platform
Positioning where
of the MoCap thecameras
system device in
was placed
relation with
to the the passive
device. Cameras reflective
are high-
markers attached to the
lighted with red circles. corners.

Figure 11. IMU test and calibration fixture developed for the characterisation.

The procedure for the data capture was as follows:

1. The cameras’ FPS was set to be the same as the sampling frequency of the IMU (100
Hz).
2. The
Figure 11. device was
IMU test andconfigured with the
calibration fixture time and
developed the characterisation.
for the IMU sampling frequency via the
Figure 11. IMU test and calibration fixture developed for the characterisation.
NFC Android application developed. Then, after placing the device in the specific
orientation
Theprocedure
The forfor
procedure the test,
forthe a start
thedata
data datawas
capture
capture recording command was sent to it.
wasasasfollows:
follows:
1.1. Thecameras’
The cameras’FPS
FPSwaswas
setset
to to
be be
thethe same
same as the
as the sampling
sampling frequency
frequency of theofIMU
the (100
IMUHz).
(100
2. Hz).device was configured with the time and the IMU sampling frequency via the
The
2. The device
NFC Androidwas configured
application with the time
developed. Then,and theplacing
after IMU sampling
the devicefrequency via the
in the specific
NFC Android
orientation application
for the developed.
test, a start Then,command
data recording after placing
wasthe
sentdevice
to it. in the specific
3. orientation
In the Motiveforsoftware,
the test, a astart data recording
recording command
was initiated, and was
the sent
timetowasit. noted to be
correlated with the data from the device.
4. The pendulum was pulled and then set free to oscillate until coming to a rest. This
period was recorded in the Motive software and inside the device.
5. The data from the device were downloaded for comparison.
 3. In the Motive software, a recording was initiated, and the time was noted to be cor-
related with the data from the device.
Sensors 2021, 21, 4649 4. 12 of 19
The pendulum was pulled and then set free to oscillate until coming to a rest. This
period was recorded in the Motive software and inside the device.
5. The data from the device were downloaded for comparison.
Data collected
Data collectedfromfromthethe
Motive
Motivesystem
systemare arethe
theposition
positionof of the
the object observa-
object under observation
tioninin
thethe Earth-fixed
Earth-fixed coordination
coordination frame,
frame, andanddatadata collected
collected fromfrom the sensor
the IMU IMU sensor are
are accelera-
accelerations
tions in theinX,the X, Y,Zand
Y, and Z directions
directions in the in the sensor-fixed
sensor-fixed coordination
coordination frame.frame. To com-
To compare these
paredata, alldata,
these of them must
all of them be must
mapped to a sametocoordination
be mapped frame and
a same coordination theyand
frame must represent
they must a
same physical
represent a same quantity. Therefore,Therefore,
physical quantity. all collectedall data were data
collected mapped weretomapped
a third coordination
to a third
frame, body-fixed
coordination frame, andframe,
frame, body-fixed then, aand second
then,derivation of data collected
a second derivation of datafrom Motive
collected
fromsystem hassystem
Motive been used as acceleration
has been measured by
used as acceleration Motive by
measured system.
Motive system.
After
After thethe data
data collection,
collection, both
both thethedatadata exported
exported from
from thethe Motive
Motive software
software and
and thethe
data downloaded from ® [50]. Both time series
data downloaded from thethe device
device were
were imported
imported into
into MATLAB
MATLAB ® [50]. Both time series
werewere synchronised
synchronised and and
thenthen compared
compared against
against each
each other,
other, as as seen
seen onon Figure
Figure 12.12.

Figure 12. Time

Figure seriesseries
12. Time of acceleration axis captured
of acceleration by the AquaBit
axis captured by the device anddevice
AquaBit the Motive
and MoCap
the Motive
system.
MoCap system.

TheThe
accuracy
accuracyof test results
of test hashas
results been
beenlimited by by
limited thethe
time synchronisation
time synchronisation forfor
AquaBit
AquaBit
andand
thethe
MoCap.
MoCap.The The
maximum
maximum errorerror
for theforsynchronisation
the synchronisationis 1 s. is
Additionally, the res-the
1 s. Additionally,
olution of time
resolution of clock in theinAquaBit
time clock the AquaBitusedused
for the
for test waswas
the test 4 ms. TheThe
4 ms. sampling
samplingfrequency
frequency
used for both the AquaBit IMU and the MoCap system was 104 Hz. The
used for both the AquaBit IMU and the MoCap system was 104 Hz. The accuracy achieved accuracy achieved
by by
calibrating thethe
calibrating MoCap
MoCap system
system forfor
thethe
experiments
experiments had a mean
had a mean error of of
error 0.60.6
mm.mm.
TheThe
results logged
results logged fromfromthisthis
testtest
cancanthen
thenbe be
used to to
used calibrate
calibrate thethe
IMUIMU using thethe
using
method
method explained
explained in Kim
in Kim andand Golnaraghi
Golnaraghi [51].
[51].

3.1.2.
3.1.2. Pressure
Pressure Sensor
Sensor
For the pressure
For the pressure sensorsensorcharacterisation,
characterisation,thethedevice
device was
was placed
placed in
inaapressure
pressurevessel
vesseland
anda adigital pressure
digital pressure gage
gage(MTI DPGA12,
(MTI DPGA12, Dwyer
Dwyer Instruments, Inc.,
Instruments, Michigan
Inc., Michigan City, IN,IN)
City, USA)
monitored the pressure inside the vessel. The air pressure inside the vessel
monitored the pressure inside the vessel. The air pressure inside the vessel was increased was increased
in intervals
in intervals of 5ofpsi
5 psi (approximately
(approximately 34.5
34.5 kPa)kPa)
eacheach 3 min,
3 min, upup
to to
30 30
psipsi (20.7
(20.7 kPa)—which
kPa)—which
is approximately equivalent to a depth of 20 m. The pressure was
is approximately equivalent to a depth of 20 m. The pressure was set manually, and set manually, and
thethe
sensor system was set to log the
sensor system was set to log the pressure. pressure.
In Figure 13, the blue line represents the pressure values measured by the gage, and
the red line represents the measurements of AquaBit pressure sensor. As we can see, the
difference between these two measurements is very low (max error is less than 6.9 Pa) in
various pressures.
 In Figure 13, the blue line represents the pressure values measured by the gage, and
the red line represents the measurements of AquaBit pressure sensor. As we can see, the
Sensors 2021, 21, 4649
difference between these two measurements is very low (max error is less than 6.9 13
Pa)of in
19

various pressures.

Figure
Figure 13.
13. Pressure
Pressure sensor
sensor test
test in
in air
air result.
result.

In Figure
In Figure 14, the result
result ofof the
thepressure
pressuretest
testininwater
waterisispresented.
presented.AsAsshown,
shown, the
thediffer-
dif-
ence between
ference betweenthethe
blue lines,
blue which
lines, whichrepresents
representsthethepressure
pressuremeasurements
measurements using a pressure
using a pres-
gage,gage,
sure and the
andorange line, which
the orange represents
line, which the measurements
represents the measurements using the
usingpressure sensor
the pressure
on AquaBit,
sensor is very small.
on AquaBit, is veryThe maxThe
small. error
maxis 0.5 psiis(3.4
error 0.5Pa),
psi which approximately
(3.4 Pa), simulates
which approximately
5 m depth5 in
simulates sea water.
m depth in seaItwater.
is worthIt isnoting that measurements
worth noting that measurementsusing using
the pressure gage
the pressure
have have
gage been been
done done
only when the pressure
only when insideinside
the pressure the pressure vesselvessel
the pressure was stable. The spike
was stable. The
Sensors 2021, 21, x FOR
in PEER
the REVIEW
pressure measured by AquaBit between 200 s to 400 s is the normal overshoot14 of that
20
spike in the pressure measured by AquaBit between 200 s to 400 s is the normal overshoot
happened
that during
happened adjusting
during the pressure.
adjusting The setup
the pressure. used for
The setup this
used test
for is shown
this in Figure
test is shown 15.
in Fig-
ure 15.

Figure 14. Pressure sensor test in water result.

Figure 14. Pressure sensor test in water result.
Sensors 2021, 21, 4649 14 of 19

Figure 14. Pressure sensor test in water result.

Figure15.
Figure 15.Test
Testsetup
setupfor
forthe
thepressure
pressuresensor
sensortest
testininwater.
water.

3.1.3.
Sensors 2021, 21, x FOR PEER REVIEW3.1.3.Light
LightSensor
Sensor 15 of 20
To
Tocharacterise
characterisethe
thelight
lightsensor,
sensor,the
thedevice
devicewas
wasplaced
placedinindifferent
differentnatural
naturallight
lightcon-
con-
ditions with a lux meter. The value read by the lux meter was manually
ditions with a lux meter. The value read by the lux meter was manually noted, and noted, and
the
the device
device recorded
recorded the the
lux lux measurement
measurement readread byinternal
by the the internal
light light sensor,
sensor, as shown
as shown in
in Figure
Figure 16 below.
16 below.

Figure16.
Figure 16.Light
Lightmeasurement
measurementcalibration
calibrationofofAquaBit.
AquaBit.

3.2.
3.2.Power
PowerConsumption
Consumption
Power
Power consumption
consumption of of the
theAquaBit
AquaBithas
hasbeen
beenmeasured
measuredusing
usingaapower
poweranalyser
analyser
(N6705B
(N6705BDC DCPower
PowerAnalyser)
Analyser)toto
figure outout
figure how
howlong thethe
long unitunit
cancan
do the sampling
do the using
sampling its
using
internal battery. As is shown in Figure 17, the power analyser was on the battery emulator
its internal battery. As is shown in Figure 17, the power analyser was on the battery emu-
mode and was
lator mode andconnected to the battery
was connected port ofport
to the battery the unit.
of the unit.
 Figure 16. Light measurement calibration of AquaBit.

3.2. Power Consumption

Power consumption of the AquaBit has been measured using a power analyser
Sensors 2021, 21, 4649 (N6705B DC Power Analyser) to figure out how long the unit can do the sampling 15 using
of 19
its internal battery. As is shown in Figure 17, the power analyser was on the battery emu-
lator mode and was connected to the battery port of the unit.

Figure 17. Power consumption measurement of battery lifetime using a DC power analyser.
Figure 17. Power consumption measurement of battery lifetime using a DC power analyser.

Sensors 2021, 21, x FOR PEER REVIEW Thepower

The powerconsumption
consumptionofofAquaBit
AquaBitisisvery
verylow
low(~1
(~1mA)
mA)forformost
mostofofthe
thetime,
time,
16 20since
of since
theunit
the unitisisininsleep
sleepmode
modeandandititgoes
goestotohigher
highervalues
valuesduring
duringthe
thesampling,
sampling,asasisisshown
showninin
Figure18.
Figure 18.TheTheaverage
averageofofcurrent
currentand
andpower
powerconsumptions
consumptionsare are1.6
1.6mA
mAand
and5.7
5.7mW.mW.

Figure
Figure 18. AquaBit
18. AquaBit current
current andand power
power consumption
consumption during
during sampling.
sampling.

TheThe battery
battery lifetime
lifetime for for
the the AquaBit
AquaBit depends
depends on the
on the sampling
sampling frequency
frequency set set
for for
thethe
sensors. In the case shown in Figure 18, the expected autonomy is of 1 week. However, for
sensors. In the case shown in Figure 18, the expected autonomy is of 1 week. However,
most deployment cases where temperature, depth, and light measurements are taken in
for most deployment cases where temperature, depth, and light measurements are taken
bigger intervals (1 h or more) the expected lifetime can be expanded. This can be configured
in bigger intervals (1 h or more) the expected lifetime can be expanded. This can be con-
to allow for customization for each deployment and study type.
figured to allow for customization for each deployment and study type.
4. Discussion
4. Discussion
Some of the most important parameters that influence seaweed growth are water
Some of
quality, the most important
temperature, parameters
light radiation, that nutrient
water pH, influenceavailability,
seaweed growth areincidence.
and wave water
quality, temperature, light radiation, water pH, nutrient availability, and wave incidence.
For an IMTA site, it is necessary to monitor these parameters with as best resolution as
possible to maximise production and minimise environmental impact. Current commer-
cial and research seaweed monitoring technology can cover large areas, but for small-
scale, fine-resolution monitoring, no integrated solution is available to the best of our
Sensors 2021, 21, 4649 16 of 19

For an IMTA site, it is necessary to monitor these parameters with as best resolution as
possible to maximise production and minimise environmental impact. Current commercial
and research seaweed monitoring technology can cover large areas, but for small-scale,
fine-resolution monitoring, no integrated solution is available to the best of our knowledge.
This work presents a solution that integrates multiple sensor modalities into a miniaturised
package that can be deployed in multiple points of a farm so that fine-scale data can be
collected and analysed.
The AquaBit device developed embeds a pressure sensor that can measure depth
and temperature, a light sensor that measures incident radiation, and an IMU to track
water movement and wave exposure. The data measured by these sensors are then logged
internally in its memory and can be transmitted wirelessly via its NFC communication
interface or wired USB connection. The device is rechargeable, reusable, and can be
customised with different sampling frequencies to adapt to different needs.
All the sensors were characterised and calibrated in lab experiments. The results show
that the sensors chosen were capable of measuring data according to the requirements
established. Therefore, the device can be used to monitor environmental factors in a specific
place in an aquaculture site, and the user can be sure that the data for each parameter are
co-located and correlated with each other.
Since the device has an internal time keeping feature, the data collected are also dated
to the second, which enables the user to plot the data in a time series and analyse it for
trends and diel patters.
The device was waterproofed with the enclosure and the PUR rated for marine use. This
means that the device can be deployed in either seawater or freshwater aquaculture farms.
The mechanical enclosure was designed to be flexible in the type of deployment, so the
user can choose how to best deploy the device for their needs. Small eye holes on the sides
allow for threading of lines to secure the device to mooring lines or to the seaweed itself.
The backside has a pattern to provide a better surface area in case the user wants to glue the
device to the seaweed/kelp blades. This type of deployment would enable the collection of
data specifically about the wave and water motion effects on the seaweed. Since the device
is very small, it does not add to the drag or additional stress to the seaweed.
As for communication, the USB interface works as expected. It can be used to commu-
nicate with the device, download data, and reprogram the device.
For the NFC communication, an Android app was developed to communicate with
the device. It can send commands (such as wake-up, shutdown, start data collection, etc.),
change the sensors configuration and sampling frequency, and download data. The range
achieved for this wireless communication is not as big as expected. This could be because of
the very small size of the antenna (limited by the device size), the presence of metal (circuit
and battery) next to the antenna, and the unoptimized NFC reader antenna (smartphone).
However, the system read range was sufficient to enable a data download using a standard
smartphone with the required capability. A dedicated NFC reader with higher transmission
power and bigger antenna could increase the read range.
An important feature of the device that we made sure was present is the customis-
ability of the sensors sampling frequency. Especially in the case of the IMU, the sampling
frequency greatly affects the type of data that can be extracted using data analysis methods.
A high sampling frequency allows the user to detect very fast movements that would
otherwise be missed. However, this needs to be balanced against power consumption, as
the higher the sampling frequency, the higher the power consumption is and the less time
the device can be deployed due to the battery lifetime.
The device was optimised for low-power consumption, from the hardware design to
the firmware development, making sure multiple power options are available. To preserve
battery while not deployed, the device has a hardware on/off controller that can put the
system in shutdown mode or wake it up via the USB or the NFC interfaces.
Sensors 2021, 21, 4649 17 of 19

As the power test shows, the device’s battery can last for a week if the sampling
frequency is set to 1 Hz. A bigger period between samples would increase the time the
device can be deployed, and this can be customised by the user.

Future Work
The next step for this study is the deployment of the AquaBit device in a seaweed
farm. This is currently in progress at the Marine Institute research facility [52].
Further improvements for the device include the development of embedded data
analysis algorithms that can process data internally in the device and then transmit only
the analysed results, thus minimising the size of transmitted data. The functionality for
this is already present in the current version, as the microcontroller chosen has enough
processing power to do so.
Another planed improvement is the addition of wireless charging that would greatly
simplify the use and deployment of the device. Since the achieved transmission range
of the NFC/RFID communication system designed for the device is small, changing the
communication to Bluetooth may not reduce the achieved transmission range so much,
but it would greatly increase the data rate and usability of the device. However, this option
would greatly increase the power consumption of the device.

Author Contributions: Conceptualization, C.P. and B.O.; methodology, C.P., H.J., M.B. and M.E.;
hardware, C.P. and M.B.; software, C.P., H.J. and M.E.; validation, M.E. and H.J.; formal analysis,
M.E.; investigation, C.P. and M.E.; writing—original draft preparation, C.P.; writing—review and
editing, B.O., C.P., H.J. and M.E.; supervision, B.O.; project administration and funding acquisition,
B.O. All authors have read and agreed to the published version of the manuscript.
Funding: This work is part of the IMPAQT project. This project has received funding from the Euro-
pean Union’s Horizon 2020 research and innovation programme under grant agreement No. 774109.
Aspects of this work were supported in part by Science Foundation Ireland (SFI) under Grant numbers
12/RC/2289-P2 (INSIGHT), 16/RC/3835 (VISTAMILK), 13/RC/2077 (CONNECT), and 16/RC/3918
(CONFIRM), which are co-funded under the European Regional Development Fund (ERDF).
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: No data.
Acknowledgments: The authors would like to thank Omid Talebi (Tyndall National Institute) for
helping with the hardware development and review; Sanjeev Kumar (Tyndall National Institute) for
helping with the antenna impedance measurement; Matteo Menolotto (Tyndall National Institute)
for helping with the Motion Capture test setup and protocol; Shree Krishnamoothy (Tyndall National
Institute) for helping with the light sensor tests.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Chopin, T.; Buschmann, A.H.; Halling, C.; Troell, M.; Kautsky, N.; Neori, A.; Kraemer, G.P.; Zertuche-González, J.A.; Yarish,
C.; Neefus, C. Integrating Seaweeds into Marine Aquaculture Systems: A Key Toward Sustainability. J. Phycol. 2001, 37,
975–986. [CrossRef]
2. Ridler, N.; Wowchuk, M.; Robinson, B.; Barrington, K.; Chopin, T.; Robinson, S.; Page, F.; Reid, G.; Szemerda, M.; Sewuster, J.;
et al. Integrated Multi—Trophic Aquaculture (Imta): A Potential Strategic Choice for Farmers. Aquac. Econ. Manag. 2007, 11,
99–110. [CrossRef]
3. Schroeder, S.B.; Dupont, C.; Boyer, L.; Juanes, F.; Costa, M. Passive Remote Sensing Technology for Mapping Bull Kelp
(Nereocystis Luetkeana): A Review of Techniques and Regional Case Study. Glob. Ecol. Conserv. 2019, 19, e00683. [CrossRef]
4. Bennion, M.; Fisher, J.; Yesson, C.; Brodie, J. Remote Sensing of Kelp (Laminariales, Ochrophyta): Monitoring Tools and
Implications for Wild Harvesting. Rev. Fish. Sci. Aquac. 2019, 27, 127–141. [CrossRef]
5. Ottinger, M.; Clauss, K.; Kuenzer, C. Aquaculture: Relevance, Distribution, Impacts and Spatial Assessments—A Review. Ocean
Coast. Manag. 2016, 119, 244–266. [CrossRef]
Sensors 2021, 21, 4649 18 of 19

6. Meneghesso, C.; Seabra, R.; Broitman, B.R.; Wethey, D.S.; Burrows, M.T.; Chan, B.K.K.; Guy-Haim, T.; Ribeiro, P.A.; Rilov, G.;
Santos, A.M.; et al. Remotely-Sensed L4 SST Underestimates the Thermal Fingerprint of Coastal Upwelling. Remote Sens. Environ.
2020, 237, 111588. [CrossRef]
7. Woo, H.-J.; Park, K.-A. Inter-Comparisons of Daily Sea Surface Temperatures and In-Situ Temperatures in the Coastal Regions.
Remote Sens. 2020, 12, 1592. [CrossRef]
8. Brewin, R.J.W.; Smale, D.A.; Moore, P.J.; Dall’Olmo, G.; Miller, P.I.; Taylor, B.H.; Smyth, T.J.; Fishwick, J.R.; Yang, M. Evaluating
Operational AVHRR Sea Surface Temperature Data at the Coastline Using Benthic Temperature Loggers. Remote Sens. 2018, 10,
925. [CrossRef]
9. Hurd, C.L. Water Motion, Marine Macroalgal Physiology, and Production. J. Phycol. 2000, 36, 453–472. [CrossRef]
10. Visch, W.; Nylund, G.M.; Pavia, H. Growth and Biofouling in Kelp Aquaculture (Saccharina Latissima): The Effect of Location
and Wave Exposure. J. Appl. Phycol. 2020, 32, 3199–3209. [CrossRef]
11. Kregting, L.; Blight, A.J.; Elsäßer, B.; Savidge, G. The Influence of Water Motion on the Growth Rate of the Kelp Laminaria
Digitata. J. Exp. Mar. Biol. Ecol. 2016, 478, 86–95. [CrossRef]
12. Bekkby, T.; Rinde, E.; Gundersen, H.; Norderhaug, K.M.; Gitmark, J.K.; Christie, H. Length, Strength and Water Flow: Relative
Importance of Wave and Current Exposure on Morphology in Kelp Laminaria Hyperborea. Mar. Ecol. Prog. Ser. 2014, 506,
61–70. [CrossRef]
13. Burrows, M.T.; Harvey, R.; Robb, L. Wave Exposure Indices from Digital Coastlines and the Prediction of Rocky Shore Community
Structure. Mar. Ecol. Prog. Ser. 2008, 353, 1–12. [CrossRef]
14. Focht, R.C.; Shima, J.S. Acceleration Loggers Reveal Fine-Scale Heterogeneity in Wave Exposure along an Open Coast. Estuar.
Coast. Shelf Sci. 2020, 233, 106507. [CrossRef]
15. Fan, L.I.N.; Meirong, D.U.; Hui, L.I.U.; Jianguang, F.A.N.G.; Lars, A.; Zengjie, J.I.A.N.G. A Physical-Biological Coupled Ecosystem
Model for Integrated Aquaculture of Bivalve and Seaweed in Sanggou Bay. Ecol. Model. 2020, 431, 109181. [CrossRef]
16. Kerrison, P.D.; Stanley, M.S.; Edwards, M.D.; Black, K.D.; Hughes, A.D. The Cultivation of European Kelp for Bioenergy: Site and
Species Selection. Biomass Bioenergy 2015, 80, 229–242. [CrossRef]
17. El Mahrad, B.; Newton, A.; Icely, J.D.; Kacimi, I.; Abalansa, S.; Snoussi, M. Contribution of Remote Sensing Technologies to a
Holistic Coastal and Marine Environmental Management Framework: A Review. Remote Sens. 2020, 12, 2313. [CrossRef]
18. García-Poza, S.; Leandro, A.; Cotas, C.; Cotas, J.; Marques, J.C.; Pereira, L.; Gonçalves, A.M.M. The Evolution Road of Seaweed
Aquaculture: Cultivation Technologies and the Industry 4.0. Int. J. Environ. Res. Public Health 2020, 17, 6528. [CrossRef]
19. Evans, S.N.; Abdo, D.A. A Cost-Effective Technique for Measuring Relative Water Movement for Studies of Benthic Organisms.
Mar. Freshw. Res. 2010, 61, 1327–1335. [CrossRef]
20. HOBO Pendant Temperature/Light Data Logger 64K. Available online: https://www.onsetcomp.com/products/data-loggers/
ua-002-64 (accessed on 17 June 2021).
21. Lyman, T.P.; Elsmore, K.; Gaylord, B.; Byrnes, J.E.K.; Miller, L.P. Open Wave Height Logger: An Open Source Pressure Sensor
Data Logger for Wave Measurement. Limnol. Oceanogr. Methods 2020, 18, 335–345. [CrossRef]
22. Kennedy, D.; Cullen, E.; McNulty, R.; Gaffney, M.; Walsh, M.; O’Flynn, B. Marine Inertial Measurement Units: Communication,
Capabilities, and Challenges. Mar. Technol. Soc. J. 2015, 49, 56–63. [CrossRef]
23. Judge, R.; Choi, F.; Helmuth, B. Recent Advances in Data Logging for Intertidal Ecology. Front. Ecol. Evol. 2018, 6, 213. [CrossRef]
24. IButton—IButton Devices—One Wire|Maxim Integrated. Available online: https://www.maximintegrated.com/en/products/
ibutton-one-wire/ibutton.html (accessed on 22 June 2021).
25. TidbiT v2 Temperature Data Logger—UTBI-001. Available online: https://www.onsetcomp.com/products/data-loggers/utbi-
001 (accessed on 22 June 2021).
26. Knight, P.; Bird, C.; Sinclair, A.; Higham, J.; Plater, A. Testing an “IoT” Tide Gauge Network for Coastal Monitoring. IoT 2021, 2,
17–32. [CrossRef]
27. Beddows, P.A.; Mallon, E.K. Cave Pearl Data Logger: A Flexible Arduino-Based Logging Platform for Long-Term Monitoring in
Harsh Environments. Sensors 2018, 18, 530. [CrossRef]
28. LTC2955 Datasheet and Product Info|Analog Devices. Available online: https://www.analog.com/en/products/ltc2955.html#
(accessed on 27 May 2021).
29. STM32L4R5/S5—STMicroelectronics. Available online: https://www.st.com/en/microcontrollers-microprocessors/stm32l4
r5-s5.html (accessed on 27 May 2021).
30. EEMBC—CPU/MCU Performance Benchmark—CoreMark. Available online: https://www.eembc.org/coremark/scores.php
(accessed on 27 May 2021).
31. EEMBC—CPU/MCU Energy Benchmark—ULPMark. Available online: https://www.eembc.org/ulpmark/ulp-cp/scores.php
(accessed on 27 May 2021).
32. Lambo, L.; Shengli, Z.; Jun-Hong, C. Prospects and Problems of Wireless Communication for Underwater Sensor Networks.
Wirel. Commun. Mob. Comput. 2008, 8, 977–994. [CrossRef]
33. Gussen, C.M.G.; Diniz, P.S.R.; Campos, M.L.R.; Martins, W.A.; Costa, F.M.; Gois, J.N. A Survey of Underwater Wireless
Communication Technologies. JCIS 2016, 31, 242–255. [CrossRef]
34. Che, X.; Wells, I.; Dickers, G.; Kear, P.; Gong, X. Re-Evaluation of RF Electromagnetic Communication in Underwater Sensor
Networks. IEEE Commun. Mag. 2010, 48, 143–151. [CrossRef]
Sensors 2021, 21, 4649 19 of 19

35. Domingo, M.C. Magnetic Induction for Underwater Wireless Communication Networks. IEEE Trans. Antennas Propag. 2012, 60,
2929–2939. [CrossRef]
36. Peres, C.; Pigeon, M.; Rather, N.; Gawade, D.; Buckley, J.; Jafarzadeh, H.; O’Flynn, B. Theoretical Models for Underwater RFID
and the Impact of Water Salinity on the Design of Wireless Systems. Int. J. Adv. Netw. Serv. 2020, 13, 45–59.
37. Finkenzeller, K. RFID Handbook: Fundamentals and Applications in Contactless Smart Cards, Radio Frequency Identification and
Near-Field Communication; Wiley: Hoboken, NJ, USA, 2014; ISBN 978-1-119-99187-8.
38. Coskun, V.; Ozdenizci, B.; Ok, K. A Survey on Near Field Communication (NFC) Technology. Wirel. Pers. Commun. 2013, 71,
2259–2294. [CrossRef]
39. European Computer Manufacturers Association. ECMA340—Near Field Communication Interface and Protocol (NFCIP-1); European
Computer Manufacturers Association: Geneva, Switzerland, 2004.
40. STMicroelectronics ST25DV04K—Datasheet 2018. Available online: https://www.st.com/resource/en/datasheet/st25dv04k.pdf
(accessed on 1 September 2020).
41. NFC Antenna EDesignSuite—STMicroelectronics. Available online: https://eds.st.com/antenna/ (accessed on 1 September 2020).
42. STMicroelectronics AN2866—How to Design a 13.56 MHz Customized Antenna for ST25 NFC/RFID Tags 2020. Available
online: https://www.st.com/content/ccc/resource/technical/document/application_note/d9/29/ad/cc/04/7c/4c/1e/CD0
0221490.pdf/files/CD00221490.pdf/jcr:content/translations/en.CD00221490.pdf (accessed on 1 September 2020).
43. STSW-ST25SDK001—Software Development Kit for ST25 Tags and Dynamic Tags—STMicroelectronics. Available online:
https://www.st.com/en/embedded-software/stsw-st25sdk001.html (accessed on 31 May 2021).
44. PyUSB. Available online: https://pyusb.github.io/pyusb/ (accessed on 31 May 2021).
45. UM1734—STM32CubeTM USB Device Library User Manual 2019. Available online: https://www.st.com/resource/en/user_
manual/dm00108129-stm32cube-usb-device-library-stmicroelectronics.pdf (accessed on 20 January 2021).
46. IMPAQT Project. Available online: https://impaqtproject.eu/about-impaqt/ (accessed on 1 September 2020).
47. Primex 13—Specs. Available online: http://optitrack.com/cameras/primex-13/specs.html (accessed on 31 May 2021).
48. Motive—Optical Motion Capture Software. Available online: http://www.optitrack.com/software/motive/index.html (accessed
on 20 May 2021).
49. Calibration—NaturalPoint Product Documentation Ver 2.2. Available online: https://v22.wiki.optitrack.com/index.php?title=
Calibration (accessed on 20 May 2021).
50. MATLAB—MathWorks. Available online: https://www.mathworks.com/products/matlab.html (accessed on 31 May 2021).
51. Kim, A.; Golnaraghi, M.F. Initial Calibration of an Inertial Measurement Unit Using an Optical Position Tracking System. In
Proceedings of the PLANS 2004. Position Location and Navigation Symposium (IEEE Cat. No.04CH37556), Monterey, CA, USA,
26–29 April 2004; pp. 96–101.
52. Our Ocean: Our Livelihoods|Marine Institute. Available online: https://www.marine.ie/Home/site-area/areas-activity/
education-outreach/our-ocean-our-livelihoods (accessed on 31 May 2021).