2016 International Conference and Exposition on Electrical and Power Engineering (EPE 2016), 20-22 October, Iasi, Romania

IoT Enabled Aquatic Drone for Environmental

Monitoring
João Matos, Octavian Postolache
ISCTE-Instituto Universitário de Lisboa, ISCTE-IUL, Instituto de Telecomunicações,
Avenida das Forças Armadas, 1649-026 Lisboa
email: jrbms@iscte.pt; opostolache@lx.t.pt

Abstract—The article describes a monitoring systembased on thoroughly with systems. This workpresents the design and
Raspberry Pi platform and a multichannel sensing module implementation of a system that can be used for water quality
associated with water quality and air quality measurement monitoring, the system being designed especially for
parameters. Thus the temperature, conductivity, relative monitoring tasks onrivers and estuaries.The capabilities of the
humidity and gas concentration are measured as was as the system is based on the usage of unmanned surface vehicle
underwater acoustic signals. The data is stored on the memoryof (USV) that helpson maritime monitoring tasks using multi-
the drone’s computational platform, and synchronized with a drone swarms [4] for extended spatial resolution. The system
remote server database. Advanced data processing algorithms
may detect such as cleaning oil spill [5], or for general
were implemented on the server side. Additionally, a mobile
environmental condition monitoring.
application was developed to be used by people working in the
field for data visualization and statistical analysis. The Internet of Things help to solve a lot of problems in
different fields: in smart cities, IoT applications are related
Keywords—Drones; USV;Sensors; Hydrophone; Raspberry P; with parking issues, noise, traffic, illumination monitoring [6];
Streaming. emergency systems for earthquakes [7]; precision agriculture
I. INTRODUCTION applications in culture process optimization [8]. IoT are used to
deliver information from the sensors and to the actuators.
The Internet of Things is an already known term nowadays
and it is becoming bigger and bigger overtime with all sorts of The goal of the paper is to present the aquatic drone setup
sensors and systems being developed to help people ease up equipped with a Raspberry Pi that is connected to an array of
their lives. The number of connected devices continues to grow sensors for air and the waterquality monitoring. For air quality
worldwide but also the diversity and the applications in the real measurements, temperature, humidity and gas sensors were
world are immense, making it an appealing industry to work on considered. For water quality monitoring water temperature
[1].A lot of companies are working on new devices and new and conductivity sensing channels were implemented.
solutions to deal with the growth of the connected devices, it is Several tasks were carried out during the system
a massive growing industry. It possesses the power to implementation. Thus the sensors interfacing with Raspberry Pi
transform every single environment such as agriculture, platform was considered, different signal conditioning modules
transportation, manufacturing, smart houses even entire cities. were implemented. Referring software Python scripts were
Companies are working on making new devices for every developed in order to read the values from the sensors and to
possible scenario but also improving communication protocols write them to a local database. A server setup was carried out
as well as security. It is estimated that around 2015 10 billion to receive the data from the local database implemented on
devices were connected and by 2020 will be connected 20 to Raspberry Pi level using database replication, in which the
30 billion devices, as costs continue to drop and demand server database will always update whenever there are changes
continues to grow [2]. in the sensor readings. In order to be possible to access and
Rivers and estuaries are incredibly important, they are a visualize the data in an appropriate GUIanandroid application
source of fresh water and nutrients required by animals and was developed. The applications fetch the data from the server
human activity, but with the development of society and database and provide live readings for the system user.
industrial demands, the pollution of rivers have become a
II. RELATED WORK
problem to be concerned about. Food production and pollution
have been putting stress within these natural resources and it is The BioMachines Lab in Institute of Telecommunications
necessary to create systems to help maintain the good health of in ISCTE-IUL, Lisbon is developing a swarm of aquatic drones
rivers. Other factor that should not be ignored is the noise designed to perform maritime tasks in the sea based onaWi-Fi
caused by human activity around the rivers but also by the ad-hoc network architecture. Each drone is equipped with a
animals themselves, it is interesting to see the relation of these compass, GPS and a Raspberry Pi 2 unit with a Wi-Fi interface
two and see if outside noise can disrupt natural activities from and they form a distributed network (controlled by an artificial
the rivers [3]. neural network based controller) without a central
coordination, they keep broadcasting their position to the
With today’s advancements in technology, it became drones in their neighborhood every second. The neural network
possible to address a larger number of problems regarding performs actions depending on what it receives, the sensor
accessibility and also monitoring and interacting more

978-1-5090-6129-7/16/$31.00 ©2016 European Union

598
readings feed the neural network and then it will control the
drone based on the information received [9].
Reported research at the International Conference on
Circuit, Power and Computing Technologies (ICCPCT) 2015
consisted ofdeveloping of a static system that includes
Raspberry Pi B+ model connected to an extra sensorssuch as:
temperature sensor. The Raspberry is part of anIoT module that
will send the data to a cloud server, the cloud server then sends
the data to a web server with a user interface [10].
In the United States, a project called the Water on the Web
(WOW) gathers information from lakes and rivers across the
different states of the country, it uses a RUSS (Remote
Underwater Sampling Station) which is a floating platform
equipped with conductivity, dissolved oxygen, pH, temperature
and turbidity sensors and it uses cellular networks to transmit
the data to the website [11].
This project resembles in a way to all of these three projects
and will sort of connect them together, it is not as thorough as Fig. 1 - Hardware System Architecture
some of them but targets the main subjects.
Drones have come a long way and have all sorts of
applications and come in different shapes, UAVs (aerial),
USVs (water surface), UUVs (underwater) and UGVs (on the
ground) [12][13]. The UAVs are perhaps the most common
with their vast application in military and the continuous
investments in that way [14]. Nowadays the drone market is
targeting more domestic and commercial usage like helping
weather forecasting [15], agriculture, movie industry, real
estate, local law enforcements, the construction business [16],
and much more that can be explored. All of this presents a new
large potential to dramatically change people’s lives, whole
industries and also introducing more privacy concerns and
security [17].
III. SYSTEM DESCRIPTION
The developed system based on Raspberry pi platform is
characterized by a hardware and software parts. The hardware
part of the system is represented in the Fig. 1:
The analog sensors are connected to the Raspberry Pi to an
ADC board based on ADS1015. The hydrophone channel
Fig. 2 - Software System Architecture
requires special analog to digital conversion and interfacing
capabilities thus an USB Audio Card was employed and The audio streaming is handled by aWowzastreaming
connected to a Raspberry USB port. service [18] installed in the server. The service Wowza
The software component part and the hardware software Streaming Engine facilitates the streaming from the Raspberry
relations are presented in Fig. 2. Pi to the server and from the server to the mobile application.
As it is presented in the Fig. 2 the Raspberry has a local As for communication is concerned, a 3G/4G connection is
database that is fetching the data from the sensors, expressed used as it is more viable due to the distances with the drone,
by water and air measured parameters and the respective being easier to provide internet connection to the Raspberry Pi
timestamps. The local database implemented at the Raspberry via a cellular receiver.
pi level is configured to be replicated at the remote server’s
side database, of which acts like a slave and copies the data IV. HARDWARE AND CONNECTIONS
from the master 1:1, every change done in the master is In this section, thedetails of the used hardwareare provided
replicated by the slave. For the server side a set of PHP scripts focussing in acquisition modules, sensors and processing units.
were developed, the scrips being accessed by the mobile
application that retrieve the sensors’ stored data. A. Acquisition modules
Since the Raspberry Pi does not includes analog inputs
additional hardware is required to acquire the signals from
analog sensors, An ADC module (Adafruit model)

599
characterized by I2C communication interface was connected local database and replicates to a remote server and also
directly to the Raspberry through the GPIO ports with proper streams the audio received by the hydrophone.
configuration. The ADC board is powered by 5V and has 4
analog input (A0-A4) that are used by the analog sensors.Using Several characteristics of Raspberry Pi are: possesses an
the python library provided by Adafruit the acquisition control ARM CPU, graphics capabilities, USB ports, Ethernet port,
is carried out. Using the developed scrips,the system HDMI port, programmable ports (GPIO) and a combined
startsgetting information from the sensors’ channels. audio and composite video jack. Its characteristics made it
The Plantronics USB audio adapter features a C-Media ideal for this application. Thus it makes it ideal to use in the
chipset which does not required any additional drivers presents environment of this project, it is possible to connect all the
full support for arm-based Linux distributions, making it ideal sensors needed and to communicate with a server to keep
to work with a Raspberry Pi. It allows the acquisition the uploading data. It also does not require much power (5V, ~1A),
signals delivered by hydrophone measurement channel. allowing it to be attached to a rather small battery for mobile
purposes [23][24]. The operating system selected was the
B. Sensing and conditioning circuits Raspbian, the official Raspberry Linux distribution, as it offers
The LM35 is a compact temperature sensor that can be official support for specific Raspberry functionalities and not
powered by 5V; it provides an accuracy of 0.5ºC for the -55ºC disregarding the advantages of having a Linux kernel.
to 150ºC specified measurement range. The sensor is already V. SOFTWARE
calibrated directly in Celsius and it provides an output voltage
linearly proportional to the centigrade temperature [20]. The software component of the system includes embedded
software and server software several details regarding backend
The temperature sensor is powered with 5V from the and frontend functionality within the Raspberry Pi, databases,
Raspberry and the output is connected directly to the ADC remote server and the mobile application are described.
module without additional conditioning required. The
implemented voltage to temperature conversion is given by: A. Embedded platform software
Several scripts were described for the Raspberry Pi. The
Temp °C =Vout·100 (1) developed software allows the system to work flawlessly. The
used software technologieswere Python, MySQL and FFmpeg.
where Vout is the output voltage of the LM35.
A Python scripts were developed for data acquisition thus
This humidity sensor provides voltage values that might be Raspberry is able to obtain the data from the measurement
converted in relative humidity (RH) values [21]. The channels including sensors. The Adafruit’s ADC library was
implemented relation to extract the RH values is used in this case.

Vout
The sensor data is uploaded to the local MySQL database
-0.16 installed in the Raspberry pi SDcard.. The script starts by
RH % =
Vsupply (2) executing a thread for each sensor with a user inputted reading
0.0062 interval; it then keeps reading the data from the ADC module
where Vout is the output voltage of the sensor and Vsupply is every X seconds and uploading to the database at the same
5V. time. At this stage the script only ends with a SIGINT signal
The TGS800 gas sensor from Figaro is recommended for sent from the keyboard combination CTRL+C which terminate
air quality control measuring general air contaminants such as: all the existing threads.
carbon dioxide, ammonia, methane, ethanol and hydrogen The MySQL database has a straightforward design, a table
through their gas concentrations in ppm. The sensor is is created for each sensor and their fields are: id, value and
characterized by a heater; the applied voltage VH on the heater timestamp. This allows the data storage associated with every
is responsible for heating up the sensor making it sensitive to sensor reading and the respective time of when the reading
the pollution compounds. An additional circuit includes the action was carried out.
sensor and a reference resistance. The circuit is powered by 5V
and the output pin connects to the ADC board [22]. To be able to stream the audio from the hydrophone, the
FFmpeg software was installed to allow a streaming channel to
The ADC module presents a limitation due to being only be made to the server. The FFmpeg receives the audio from the
12-bit, both humidity and gas sensors easily saturated with the ALSA input microphone channel, which is the general
increase of output voltage which would reach the limit of software in Linux to control all the audio/video devices and
4.096V from the ADC. In order to solve this problem, a voltage provides drivers in order for the devices to work in Linux. It
divider was applied to attenuate the output voltage in half and then captures the microphone data and creates an RTMP
then compensated back at a software level. channel that publishes the stream to the remote server.
C. Computational Platform B. Server side software
The Raspberry Pi 3 is the core of the system; it deals with The server houses a couple of PHP scripts to allow the
all the important data input and output. It receives data from all mobile application to access the sensors data. A total of 4
the analog sensors and the hydrophone and saves the data in a scripts were developed, three of them will return all of the
saved data from each sensor in a JSON format and the last

600
script will return the latest recorded value for each sensor,
depending on the “GET” variable is passed by the mobile
application. If a GET string “temp” is passed, the script returns
the latest temperature value recorded the same goes for the
“hum” and “gas” strings for their respective sensors. The
purpose of this last script is to allow the mobile application to
keep getting the latest value when it desires to show live
readings with a certain interval.
In order of the streaming to work, it is also necessary to
configure the server to act as the distributer of the audio
stream. It receives a stream publication from the FFmpeg
software in the Raspberry Pi and the mobile application will
subscribe to that same stream via the same RTMP link. For all
this to work, a streaming service from Wowza, the Wowza
Streaming Engine was installed in the server to facilitate the
streaming process. Wowza generates a link where it is allowed
to publish streams with almost any audio/video streaming
protocol and at the same time allows for subscriptions that lets
a client software to play the audio/video, in this case, the
mobile application. Fig. 4 – Gas Sensor Output Voltage (VG) evolution according with indoor
gases’ concentrations conditions
C. Android Application
The android application is at this moment in an early stage, In this case the ethanol concentration was put really close to
it has a very simple UI that shows the sensors latest reading the sensor, explaining the high peak value. As the ethanol is
values in Card Views. It reads the data through connecting to removed from the air, the sensor takes time to adjust to normal
the remote server via HTTP and executing the PHP scripts that levels as the rest of the gas inside the sensor evaporates.
will return the data. Regarding the pollution event detection capabilities of the
VI. PRELIMINARY RESULTS AND DISCUSSION implemented system an ethanol pollution event was induced
the broad band gas sensor response being presented in Fig. 5.
The implemented system was partially tested by
components before installation on a USV. The sensing Several tests were also performed using the hydrophone
channels were tested one by one in the laboratory conditions. mobile software component being designed and implemented
to extract information about underwater audio spectral
For the temperature and relative measurement case the components. In Fig. 6 is presented the GUI of the underwater
acquired and processed values are presented in the Fig. 3. acoustic spectrum.
As expected, the values do not vary too much as it is an
indoor measurement during the night, temperature was always
around ~26ºC and humidity always around ~39%.
As for the gas concentrations, the following figure shows
the sensor being turned on for a couple of hours and it can be
seen that the sensor takes around 2-3 hours to calibrate.

Fig. 5- Gas Sensor Output Voltage (VG) evolution with the presence of
ethanol

Fig. 3 – Temperature(T) and Relative Humidity (RH) Measurements in

Indoors conditions

601
 VII. CONCLUSIONS AND FUTURE WORK
The goal of this paper is to provide information about
UAV prototype implementation for water and air quality
monitoring. The preliminary data underline the capabilities of
the implemented system to provide information about water
and air quality for an extended area. Future work will be
focussed on mobile application test and improvement
regarding functionalities and robustness. On the software side,
the mobile application is also developed to provide data
analysis and audio streaming. Additionally,field tests related
to water and air quality monitoring together the
implementation of measurement channel calibration
procedures that will assure the appropriate measurement
accuracy for the system will be carried out.
REFERENCES
[1] O'Brien, H. Michael, 2016, The Internet of Things. Journal of Internet
Law. 2016. Vol. 19, no. 12, p. 1-20.
[2] Bughin, Jacques, Chui, Michael and Manyika, James, 2015, An
executive's guide to the Internet of Things. Mckinsey Quarterly. 2015.
No. 4, p. 92-101.
Fig. 6–Spectral analysis of audio signal provided by hydrophone implemented [3] Wang, Biao, Lu, Shi-qiang, Lin, Wei-qing, Yang, Yi-fan and Wang, Dao-
in the mobile APP zeng, 2016, Water quality model with multiform of N/P transport and
transformation in the Yangtze River Estuary. Journal of Hydrodynamics.
2016. Vol. 28, no. 3, p. 423-430.
In the mobile application case, the GUI associated with on-
[4] Velez, Fernando J., Nadziejko, Aleksandra, Christensen, Anders Lyhne,
line monitoring of air and water quality parameters is Oliveira, Sancho, Rodrigues, Tiago, Costa, Vasco, Duarte, Miguel, Silva,
presented in Fig. 7. This is the main screen of the mobile Fernando and Gomes, Jorge, 2015, Wireless Sensor and Networking
application, where the measuredvalues delivered by sensors Technologies for Swarms of Aquatic Surface Drones. 2015 IEEE 82nd
are displayed “in cards” and are updated in real-time. The Vehicular Technology Conference (VTC2015-Fall). 2015.
DOI 10.1109/vtcfall.2015.7391193. Institute of Electrical & Electronics
capabilities of the implemented system on pollution event Engineers (IEEE)
detection were also tested. Thus an ethanol pollution event [5] Zahugi, Emaad Mohamed H., Shanta, Mohamed M. and Prasad, T.V.,
was induced and the broad band gas sensor response is 2012, Design of Multi-Robot System for Cleaning up Marine Oil Spill.
represented in Fig. 5. International Journal of Advanced Information Technology. 2012. Vol. 2,
no. 4.
The mobile application receives the data from the remote [6] Burange, Anup W. and Misalkar, Harshal D., 2015, Review of Internet of
server via the implemented PHP scripts.Based on the remote Things in development of smart cities with data management &
server scripts the signal data after processing can be privacy. 2015 International Conference on Advances in Computer
Engineering and Applications. 2015. DOI 10.1109/icacea.2015.7164693.
transferred to the Raspberry Pi. Institute of Electrical & Electronics Engineers (IEEE)
[7] Spalazzi, Luca, Taccari, Gilberto and Bernardini, Andrea, 2014, An
Internet of Things ontology for earthquake emergency evaluation and
response. 2014 International Conference on Collaboration Technologies
and Systems (CTS). 2014. DOI 10.1109/cts.2014.6867619. Institute of
Electrical & Electronics Engineers (IEEE)
[8] Ye, Jiuyan, Chen, Bin, Liu, Qingfeng and Fang, Yu, 2013, A precision
agriculture management system based on Internet of Things and WebGIS.
2013 21st International Conference on Geoinformatics. 2013.
DOI 10.1109/geoinformatics.2013.6626173. Institute of Electrical &
Electronics Engineers (IEEE)
[9] Duarte, Miguel, Costa, Vasco, Gomes, Jorge, Rodrigues, Tiago, Silva,
Fernando, Oliveira, Sancho Moura and Christensen, Anders Lyhne, 2016,
Evolution of Collective Behaviours for a Real Swarm of Aquatic Surface
Robots. PLOS ONE. 2016. Vol. 11, no. 3, p. e0151834.
DOI 10.1371/journal.pone.0151834. Public Library of Science (PLoS)
[10] Vijayakumar, N and Ramya, R, 2015, The real time monitoring of water
quality in IoT environment. 2015 International Conference on Circuits,
Power and Computing Technologies [ICCPCT-2015]. 2015.
DOI 10.1109/iccpct.2015.7159459. Institute of Electrical & Electronics
Engineers (IEEE)
[11] Waterontheweb.org, Water on the Web | Data |. [online]. [Accessed 13
July 2016]. Available from:
http://www.waterontheweb.org/data/index.html
[12] DiNardo, P.B. and Andrews, S.B., [no date], Unmanned systems
Fig. 7–Air & Water Quality Mobile APPs GUI initiative: supporting immediate warfighter needs and army

602
 transformation. ROMAN 2005. IEEE International Workshop on Robot [17] Rao, Bharat, Gopi, Ashwin Goutham and Maione, Romana, 2016, The
and Human Interactive Communication, 2005.. societal impact of commercial drones. Technology in Society. 2016.
DOI 10.1109/roman.2005.1513765. Institute of Electrical & Electronics Vol. 45, p. 83-90. DOI 10.1016/j.techsoc.2016.02.009. Elsevier BV
Engineers (IEEE) [18] Media Server Software | Wowza Streaming Engine. Wowza.com [online]
[13] Patterson, Mark R. and Patterson, Susan J., 2010, Unmanned systems: An [19] ADS1015 / ADS1115 | Raspberry Pi Analog to Digital Converters |
emerging threat to waterside security: Bad robots are coming. 2010 Adafruit Learning System.Learn.adafruit.com [online]
International WaterSide Security Conference. 2010.
[20] LM35 Precision Centigrade Temperature Sensors. Ti.com [online]
DOI 10.1109/wssc.2010.5730271. Institute of Electrical & Electronics
Engineers (IEEE) [21] HIH-4000 Series Humidity Sensors. farnell.com [online]
[14] Magnuson, Stew, 2016, Military beefs up research into swarming drones. [22] Figaro Gas Sensors. jens.files.wordpress.com [online]
National Defense. 2016. Vol. 100, no. 748, p. 22. [23] Raspberry Pi, What is a Raspberry Pi? [online]. [Accessed 12 July 2016].
[15] Elston, Jack, Argrow, Brian, Stachura, Maciej, Weibel, Doug, Lawrence, Available from: https://www.raspberrypi.org/help/what-is-a-raspberry-pi/
Dale and Pope, David, 2015, Overview of Small Fixed-Wing Unmanned [24] Raspberrypi.org, Raspberry Pi Hardware - Raspberry Pi Documentation.
Aircraft for Meteorological Sampling. Journal of Atmospheric & Oceanic [online]. [Accessed 12 July 2016]. Available from:
Technology. 2015. Vol. 32, no. 1, p. 97-115. https://www.raspberrypi.org/documentation/hardware/raspberrypi/READ
[16] Unmanned Aircraft Systems Overview, 2016. Congressional Digest, ME.md
Vol. 95, no. 6, p. 2-32.

603