Proposed open source architecture for Long Range monitoring.

The
case study of cattle tracking at Pogoniani.

Nicholas Zinas Sotirios Kontogiannis George Kokkonis

TEKMON Geomatics LLP Dept. of Mathematics Dept. of Business Administration
Ioannina, Greece University of Ioannina Techn. Edu. Inst. of Western
[email protected] Ioannina, Greece Macedonia, Grevena, Greece
[email protected] [email protected]

Stavros Valsamidis Ioannis Kazanidis

Dept. of Accounting and Finance, Dept. of Accounting and Finance,
Techn. Edu. Inst. of East Macedonia, Techn. Edu. Inst. of East Macedonia,
Kavala, Greece Kavala, Greece
[email protected] [email protected]

ABSTRACT
Wireless communications play a significant role in the spread of 1 INTRODUCTION
the Internet of Things. Power consumption, Long range coverage,
bandwidth and security are critical factors for the blossom of each
Over the last years, the Internet of Things (IoT) is gaining more
wireless technology. This paper presents authors experimentation
and more attention and appreciation. Machine to Machine
and implementation journey to open source IoT for animal
wireless communication protocols and services are being
tracking. Focusing on cows, it describes the LoRaWAN
developed. Actuators and smart equipment interoperate with
architecture for long range communications and analyses the high
sensors and the Internet cloud. The technology that will endorse
level system architecture for cattle tracking. Furthermore, a
smart houses, cities, and mobile wearable instrumentation is on
software application and protocol has been developed and
the way.
presented.
Similarly in agriculture, GIS, facility and monitoring systems are
immersed with sensors and smart algorithms. Such algorithms
CCS CONCEPTS assist, predict and automatically drive decisions and actuators.
Applied computing → Computers in other domains → Experimental of these attempt efforts are self driven tractors [20],
Agriculture sprinkling drones, animal tracking systems and autonomous
irrigation systems [25] are some of the appliances put to test
KEYWORDS nowadays [21, 22]. Focusing on cow tracking the authors shall
Cow Tracking, IoT, LoRa, LoRaWAN, sensors and application present their adventure to this field of IoT applied research in
protocols, incident response system and architecture, monitoring agriculture.
systems 1 Existing cow tracking systems use LTE networks to upload
sensory and GPS data to their application services [7, 8]. Since
power consumption, and range coverage are the main factures that
determine the technology used, and since 2G/3G communication
transponders are power consuming (at least 140mA for data
Permission to make digital or hard copies of all or part of this work for personal or transmission), new technologies for Low Power WAN have
classroom use is granted without fee provided that copies are not made or distributed emerged. The most representative ones are LoRa and SigFox.
for profit or commercial advantage and that copies bear this notice and the full
LoRa is a long range radio protocol equipped with a MAC layer
citation on the first page. Copyrights for components of this work owned by others
than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, for robustness called LoRaWAN. Lora developed by SemTech
or republish, to post on servers or to redistribute to lists, requires prior specific and supported by the LoRa alliance [1, 2, 19, 26], can cover areas
permission and/or a fee. Request permissions from [email protected]. up to 15-20Km. Moreover, authors’ experimentation overseas
PCI 2017, September 28–30, 2017, Larissa, Greece
© 2017 Association for Computing Machinery.
reached communication distances up to 36Km.
ACM ISBN 978-1-4503-5355-7/17/09…$15.00 SigFox from on the other hand is a French company that builds
https://doi.org/10.1145/3139367.3139437 low throughput transponders and protocols [11]. SigFox motes
PCI 2017, Sep 2017, Larisa, Greece N. Zinas et al.

use commodity MSK radios which are relatively inexpensive. infrastructure of smart cities, due to the high throughput achieves.
However, SigFox is setting network operators to pay royalties on However 433MHz is much more efficient under tree clusters
reselling its technology services to customers. In many cases, the giving a better distance with less max throughput than 868MHz
company actually deploys the network and acts as the network devices.
operator. SigFox transponders operate at 902MHz, and manage For IoT conectivity data protection and LoRa scalability purposes,
very low throughput of less than 1Kb/s over wide are distance of a communication protocol and architecture has been developed
up to 30-60Km. Cow tracking solutions such as [9, 10] follow over LoRa physical layer. LoRaWAN is the network protocol that
SigFox LPWAN implementation. LoRa technology from the other utilizes LoRa as the physical layer [2, 24, 26, 19]. LoRaWAN
hand is free to use and for that purpose it utilizes the ISM bands normally is used for data rates for devices which can be from
for data transmission. It achieves higher throughput but less 300Bps up to 5.5Kbps. With the exception of two high speed
coverage than SigFox [9, 23]. channels of 11Kbps and 50Kbps (using FSK modulation) [24]. Tx
of LoRa is between 14-23dBm while min reception signal
As cattle have the freedom of movement, they can move up to sensitivity is up to -130dBm [2, 24], bellow white noise levels.
several miles every day. The communication system architecture LoraWAN Gateways are the intermediate connection systems
that will be chosen should manage to cover an area up to 50 between LoRa IoT devices and the Internet. LoraWAN gateways
square kilometers. As this communication is based on batteries the are constructed with the use of Lora concentrators. They can listen
power consumption is also a critical factor for the correct choice on up to 8 frequencies and can receive concurrently two different
of the wireless communication protocol. spreading factor devices (SF) on the same frequency. In cases of
For the purposes of an open source operator free cow track collision, packets with the strongest signal get decoded first. In
implementation authors turned to LoRa. That is because LoRa cases of link budget decrease, gateway automatically adapts its
fulfills low cost, long range coverage, low power consumption data rate using its internal ADR mechanism provided by the lora
and the security that is needed in applications as livestock service, that changes the connected equipment SF. The ADR
monitoring. In rural areas with line of sight, the LoRaWAN MAC mechanism works like this: If you get a packet from a node with a
protocol manages to scale for up to 20-30 devices. For security, low or high SNR value then the gateway adjusts the SF
the LoRaWAN technology enforces the AES 128 encryption. As accordingly. SF parameter adjusts the packet's time on air interval.
for power consumption the modules consume 40-45 mA at 3.3 for example SF9 is 4 times slower than SF7 in terms of encoding
Volt transmits and the receivers have -140 dBm sensitivity. The crisps per second. This way the data rate can be adjusted. Time on
LoRa modules in EU operate at 433 and 868MHz frequency air is proportional to the SF value used. SF are typically driven by
bands. Nevertheless, only the 868MHz implementation supports the application server using a dBm reception table that ranges
the LoRaWAN architecture. Since 433MHz band is much more from SF=7 if the received packet SNR=-7.5dB (5Kbit/s-56ms for
suitable for rural areas, authors are experimenting into the 10 bytes time on air) to SF=12 if the received packet SNR=-20dB
implementation of LoRaWAN capable gateways and services (293Bit/s – 1483ms for 10bytes time on air) [24, 20]. In cases of
under the 433MHz band. The main problem with LoRa is its congestion the installation of more LoRa gateways or the
centralizes architecture setup, described in the following section. reduction of IoT equipment transmission period can mediate the
problem.
2 LORA CAPABILITIES AND LORAWAN LoRaWAN is the logical and link layer communication protocol
ARCHITECTURE over the LoRa physical layer. LoRaWAN also includes a specific
network architecture that includes the following components:

LoRa is extremely low power, low BW technology. It is 1. LoRaWAN capable sensors: This is the sensors equipment
absolutely not comparable with Wi-Fi standards. It is much closer
that is controlled by a microprocessor unit of at least 8bit
to the mobile internet standards as 3G/4G apart from low power
MCU, at least 32KB of volatile memory and 1KB of
variants which are in the process of development (5G). It is an
EEPROM for storage of configuration parameters and keys
ideal technology for IoT cases where sensors do not transmit a lot
used by the LORAWAN authentication and data encryption
of data (class A LoRaWAN mode [24, 26]).
process [3, 4]. However, the commonly used LoRaWAN
The Link budget for Lora is 154dB according to Semtech, which
is much higher that LTE Cat-1 link budget of 130dB. For a LOS enabled microprocessors are 32bit capable with at least
connection between sender and receiver the biggest distance for 256KB of flash and 32KB of RAM. Representatives are the
LTE does not breach the theoretical coverage of 100Km which is Microchip LoRa motes, that use a RN2483 transceiver of
close to -131.5dB. Practically, its maximum range is close to 40mA (14dB) transmission power and controlled by a 8bit
80Km. On the other LORA can achieve a theoretical coverage of PIC microcontroller [5], the Libelium WASPmotes [6], the
30000Km using power transmissions similar to an LTE (FDD) BS SemTech SX127x kits that include SX1272/3/6 transceivers
transponder EIRP (average of 46dBm). (up to 20dBm power and 154dB link budget) and the
Lora devices operate in the license free ISM bands of 433MHz SODAQ [18] / Adafruit Feather M0 motes [4] that include an
and 868MHz with the EU sub-band duty-cycle limitation of 0.1%- RN2483 and RFM96 LoRa transponder accordingly
1% [24]. The band used in EU is 868MHz supporting the controlled by ATMEL M0 arm cortexes (up to 23dBm power
2
PCI 2017, Sep 2017, Larisa, Greece N. Zinas et al.

and 120dB link budget). In this case study implementation application was implemented by authors and queries the
the authors used the SODAQ one v2 mobile components. Mongo database in order to obtain information regarding the
2. LoRa Gateway: Gateways are the data collectors of the nodes status as well as the sensors’ payload data they carry.
LoRaWAN architecture. They are listening to 868MHz Nodes payload data are usually stored to the database and
channels for data transmission. Typically there seven send to the MQTT broker using base64 encoding for data
channels used for uplink transmission as well as one integrity purposes. It is responsibility of the CowTrack
common channel for RX2 (receiver window 2) downlink application to base64 decode the nodes’ payload data and
data reception [24, 12]. Normally a LoRaWAN gateway further decode them according to the custom Application
includes a concentrator like the IMST iC880A, which protocol used, in order to obtain the real sensory data. The
includes at two SX1273 transponders capable on listening CowTrack application protocol implemented by authors and
periodically to all transmission channels. However there are CowTrack application capabilities are described at section 4.
single channel custom made gateways capable on monitoring
only one channel. The LoRa gateway is equipped with an 3 LORAWAN LONG RANGE
external control microprocessor. In authors’ case a RPi 3 unit COMMUNICATION EXPERIMENTS
connected to the IMST concentrator via the SPI bus. This
interconnection between the concentrator and the In order to test the coverage range of the LoRaWAN modules, the
microprocessor is supported by appropriate open source authors performed some experiments in rural areas with line of
software such called gateway packet forwarder, which is sight. They send data at 50, 500, 1500 and 4000 meters away from
capable to capture node class A uplinks and forward packets from a LoRaWAN gateway that uses a 2x SX1273 IMST
to and from the LoRa application server using UDP protocol concentrator board at different rates. Then authors measured the
for uplinks and status control downlinks [24, 13, 14, 26]. signal power and the throughput at the LoRa application server.
LoRa Gateways also have internet connectivity supported by The max throughput was recorded only when the packet loss or
IP tunnels in order receive and transmit data to the LoRa packet CRC failure was lower than 3%. The mobile equipment
server module that was used was the RN2483 class A component
3. LoRa Gateway Bridge: Upon packet reception to the LoRa connected to the Gateway via ABP authentication, and send data
gateway it is forwarded to a UDP service instantiated at the using different inter-frame intervals. The core results of this
LoRa server, called Gateway Bridge [12, 26]. The LoRa experimentation are depicted at Table I.
gateway bridge forwards the packets over the LoRa server
MQTT broker [15]. TABLE I. Measurements for signal power and
throughput MEASURED at the LoRa Concentrator module
4. LoRa server: This is the core application server of the
LoRaWAN architecture [12]. It supports node authentication Range (m) Signal Max throughput
via ABP and OTAA [24, 26, 2]. It utilizes services as Power (Kbit/s)-Packet
postgreSQL and Redis [17] for node keys and configuration
(dBm) loss <3%
parameters storage as well as temporary session keys and
decrypted data storage accordingly. It also supports adaptive 50 -45 45
data rate LoRa mechanism [24, 26] and is responsible for 500 -72 38
data decryption prior the received data payload reach on the
1500 -89 14
MQTT broker.
5. LoRa Application server: This is the web-based (HTTP- 4000 -104 1.6
SSL) administration and monitoring service of the
authenticated nodes per application, as well as monitoring As it was expected the signal power and the throughput decreases
interface of the connected gateways to the infrastructure [12]. as the range increases. The mobile equipment successfully
6. MongoDB service and tools: MQTT broker offers only real- managed to upload data stream at 1.6 Kbps with packet loss lower
time data snapshots. For that purpose, authors incorporated to than 3% at 4000 meters. The signal power at that distance was -
the architecture a MongoDB service [16] and implemented 104dBm.
the appropriate Python utilities for abstract data reception
from the MQTT broker and storage to MongoDB using a
separate collection per LoRaWAN application [12, 26]. 4 HIGH LEVEL SYSTEM ARCHITECTURE
7. The CowTrack application: For authors case study, inside OF COW TRACKING APPLICATION
the MongoDB there is a collection called cowtrack following
the LoRa application server name set by the administrator 4.1 Mote equipment and sensors used
when configuring keys and nodes for the application at the For the case study of long range monitoring we adopted the case
LoRa Application server web interface. The CowTrack web of cattle tracking at Pogoniani, Ioannina, Greece. For the needs of

3
PCI 2017, Sep 2017, Larisa, Greece N. Zinas et al.

this case study we implemented a web based, android application on the top of the project box. The battery can practically support
named CowTrack. The primary target of this application is to up to 35000 transmissions.
monitor the environmental conditions and GPS position of the
cows. The sensors that are used are:
Temperature sensor: Air temperature is also monitor with the 4.2 Cow tracking Application protocol
help of the digital sensor inside the SODAQ component. The
accuracy of measurement is ±0.5°C.
The data that is send each time the cattle moves with an
GPS: The Ublox Eva 7M GPS is used. Whenever the cow starts
acceleration speed of more than 10m per min is 32 bytes and is
to move the Latitude and Longitude of its position is captured
depicted at Figure 2.
with the help of the GPS sensor.
Acceleration sensor: The acceleration of the cow is constantly
measured. The acceleration sensor that is used is the LIS3DE
.When the cow starts to move with acceleration higher to 1 m/sec2
then the microcontroller measures all the sensory data and send
them to the data center through the LoRa transceiver.
Battery sensor: The life of the battery is measured with the help
of the microcontroller. The voltage of the battery is measured and
Figure 2: CowTrack application protocol structure (32 Bytes).
the data center is informed via updates.
Magnetic reed sensor: It is used for alerting the data center when
The data that are monitored and send to the LoRa application
the collar at the cattle’s neck is broken. It is used for notifying the
server via class A updates are the following:
data center if a thief is trying to cut off the transmitter from the
Latitude: It is obtained from the GPS signal and is 8 bytes long.
cattle’s neck at steal the cattle.
Longitude: It is also obtained from the GPS signal and is 8 bytes
The microcontroller that is used for the control of all the above
long
sensors is the Sodaq one v2 [18]. It is a 32 bit Arduino compatible
Altitude: It is two bytes and is given by the GPS sensor.
Atmel processor that encloses the GPS, the RN2483 LoRa
Number of Satellites: The number of satellites that are
transceiver and the Acceleration sensor. The SODAQ modules
synchronized with the GPS sensor. This value determines the
uses two antennas. A GPS pack antenna, positioned in the upper
accuracy of the GPS signal. One byte is needed for that
side of the cows bell and a LoRa molex antenna. Also a reed
measurement and at least a value of 4 to get an updated value for
sensor is placed at the belt’s clasp and covers the whole belt with
lat, long instead of a cached one.
the use of conductive thread. The reed sensor informs the system
Acceleration: For the measurement of the acceleration of the cow
whether the belt has been removed or cut.
2 bytes are used and the value is measured as meters per minute.
Temperature: The temperature in Celsium degrees is given by
the SODAQ (2 bytes long)
Battery life: The battery life is measured by the SODAQ and is 2
bytes long. It is measured in decades of uAh. Typically each
transmission (including GPS fix and sensor measurements) draws
about a value of 416 (/10) which corresponds to a current of
149.76mA. That is, each device powered by a 2500mAh battery
can support a theoretical of 60,000 measurements.
Direction: The direction of the cow’s movement is obtain by the
GPS signal and has accuracy of 2 degrees. One byte is used.
UTC Timestamp: 4 bytes are kept for timestamp from the
microcontroller’s internal clock.
NSS1: It is used to inform the data center whether the project box
is correctly attached to the cows neck. For that purpose a reed
Figure 1: Enclosure of the microcontroller, sensors and
sensor is connected to round the cow’s collar. One byte is needed
transceiver in a project box.
for this sensor value.
NSS2: 1 byte for future 8 digital or one analog sensor
All the above sensors, microcontrollers and transceivers are
measurement use.
enclosed in a project box as illustrated at Figure 1 and placed in
the bell-collar of the cow. The molex antenna is inside the plastic
box. The weight of the bell under the belt, keeps the SODAQ box
4.3 Cow tracking web Application and
on the upper side of the cow’s neck. The battery that is used for capabilities
the SODAQ power supply is a 2500 mAh Lithium battery. In All data collected from the remote sensors are stored in an open
order to extend the life of the battery a 1 watt solar panel is placed source MongoDB database. These data are managed by the

4
PCI 2017, Sep 2017, Larisa, Greece N. Zinas et al.

Cowtrack Application manager (see Figure 3). The stored fields  If a cow is starts to run with a speed greater than a threshold
are: timestamp, longitude, latitude, altitude, speed, battery life, then a notification with the cow’s GPS information is also
GPS accuracy, acceleration, cow colar status and temperature. sent by the Cowtrack application.
 If the battery life of a node is below a threshold again a
notification is sent by the Cowtrack application.
 Cowtrack application real time monitoring and raw data
MongoDB explorer/manager per LoRa application (cattle)
and subscribed node (see Figure 3).

7 CASE STUDY OF CATTLE TRACKING

The cattle tracking problem has been addressed almost a year ago
when a cow farmer reached at Tekmon’s office asking for help.
The facing problem was that his cow herd (approximately 87
cows) was hard to monitor even by his employees following his
herd. Occasionally his employees were distracted, loosing unity of
the heard that graze in scattered places. There were also times
Figure 3: Raw mongoDB data explorer of the CowTrack where some of his cows have been stolen and he wanted a
application.
tracking monitor system for both cows and employees.
The case study took place in an area called Pogoniani near the
Albanian borders. Based on the farmer’s requirements of long
range coverage area with low power transmitters of minimum
cost, the system architecture of LoRa was selected and the
LoRaWAN architecture was implemented experimentally using
two IMST gateways and two SODAQ nodes. Also the CowTrack
application and application protocol have been developed, as well
as the SODAQ mote placement around the cow’s belt.

Figure 4: Logical fence for alert notifications UI of the CowTrack

application.

Besides displaying raw data via the Cowtrack manager (see

Figure 3), the web application has also the following capabilities:

 Display of the cow’s real time position in a google maps

overlayed Cowtrack application layer , (see Figure 4). It can
map overlay the cow’s route of the last day, week or month Figure 5: Case study Cattle’s nodes route on Google maps.
(see Figure 4).
 On the Display map the shepherd can draw a logical fence Then the proof of concept tests of minimum-maximum battery
where the cattle should pasture. If the cattle passes out of the replacement time, maximum coverage area took place. The
logical fence an alert via SMS or e- mail is sent to the smart preliminary tests showed that the coverage area required could be
phone of the shepherd (see Figure 4). satisfied if at least one getaway is placed, for a number of 20-30
 If the cow’s collar is broken or removed by force than an SODAQ motes (placed into the 24% of his heard). From the cost
alert is also sent by the Cowtrack application. analysis that followed lead us to the placement result of one mote
per two cows. That is, that the coverage of one gateway was
enough to cover more than 50% of his herd and was set to be
placed for the CowTrack application tests at his stable, where

5
PCI 2017, Sep 2017, Larisa, Greece N. Zinas et al.

electrical power and ADSL internet connectivity were available. “CattleSense develops smart GPS collars to keep track of cow
The temporary placement has been performed at the farmer’s silo herds”, http://nocamels.com/2013/02/cattlesense-using-satellites-
and the experimentation of the CowTrack application took place. and-ai-for-cattle-herding/, 2013.
In this experimentation two SODAQ motes have been used into [8] I. Arbel et. al., “CattleWatch application, remote heard’s
two separate cows starting from points A (see Figure 5). The cows health, location, and theft warning”, http://www.cattle-watch.com,
2015.
by the evening should reach point B which is 5.87Km away. The
[9] Sensowave Ltd, “DigitAnimal cattle monitoring platform and
farmer was given a tablet where he was connected to Internet stepla equipment based on Sigfox”,
monitoring in real-time its two cows moving from A to B. From https://digitanimal.com/digitanimal-platform/, Spain, 2015.
the application experimentation the farmer was satisfied with the [10] CSIRO, Robotics and Autonomous Systems company,
implementation, he pointed out minor bugs and correction that “Wireless animal sensor node for tracking cows”, based on
need to be made and suggested an android application for his US20090211538 Animal management systems patent, 2009.
phone as a better monitoring tool. The porting of the CowTrack [11] SigFox company, “The Global Communications service for
web application to an Android application is currently an ongoing the Internet of Things”, https://www.sigfox.com/en, 2011.
project. [12] O. Brocaar, “The open source LoRaWAN application server,
loraserver and gateway bridge”, https://github.com/brocaar, 2015.
[13] M. Coracin, Semtech company, “The Lora network packet
8 CONCLUSIONS forwarder project”, https://github.com/Lora-net/packet_forwarder,
2013.
In summary, Authors implemented a CowTrack protocol and [14] E. Gourlauen and J. Stokking, “The ThingsNetwork gateway
application based on the LoRaWAN technology and MongoDB. packet forwarder”,
Authors managed to test their implementation at the area of https://github.com/TheThingsNetwork/packet_forwarder, 2013.
[15] MQTT mosquito, “An open source MQTT broker and tools”,
Pogoniani, Ioannina, Greece, prior to a series of LoRaWAN
https://mosquitto.org/, 2011.
technology experimentations. The experiments revealed that
[16] MongoDB, “A schema less database engine”,
authors could send data over LoRaWAN to a distance of 6Km. https://docs.mongodb.com/, 2011.
The authors managed to demonstrate successfully their CowTrack [17] RedisLabs, “Redis-An open source in-memory data structure
implementation using two LoRa modules. Furthermore, the extra store, used as a database, cache and message broker”,
capabilities of the CowTrack web application such as cow fencing https://redis.io/, 2012.
and alerts have been tested by the farmer and were given a [18] Sodaq company, “Sodaq one v2 LoRa mote”,
positive feedback. Authors set as future work the implementation http://support.sodaq.com/sodaq-one/sodaq-one/, 2016.
of a mobile application that will interoperate with the CowTrack [19] F. Adelantado, X. Vilajosana, P. Tuset-Peiro, B. Martinez, J.
web application service and modification of LoRaWAN Melia-Segui and T. Watteyene, “Understanding the Limits of
LoRaWAN”, IEEE Communications magazine, Feb. 2017.
technology band at 433MHz.
[20] C. Civelek, “Low Power Wide Area Network (LPWAN) and
Internet of Things Adaptation in Agricultural Machinery”,
Scholars Journal of Agriculture and Veterinary Sciences, Vol. 4
REFERENCES No.1, ISSN:2348-8883, 2017.
[1] V. Lorenzo, A. Zanella, and M. Zorzi. “Long-range IoT [21] R.P. Pratim, “Internet of things for smart agriculture:
technologies: The dawn of LoRa™”, Future Access Enablers of Technologies, practices and future direction”, Journal of Ambient
Ubiquitous and Intelligent Infrastructures., pp. 51-58, Springer Intelligence and Smart Environments, vol. 9, no. 4, pp. 395-420,
International Publishing, 2015. 2017.
[2] B. Dmitry, E. Khorov, and A. Lyakhov. “On the Limits of [22] H. Tasew Reda, P. Tobianto Daely, J. Kharel and S. Young
LoRaWAN Channel Access”, In Proc. of IEEE International Shin, “On the Application of IoT: Meteorological Information
Conference on Engineering and Telecommunication (EnT), Display System Based on LoRa Wireless Communication”, IETE
Moscow, Russia 2016. Technical Review, Vol. 1, No. 1.
[3] M. Kooijman, “Arduino port of the LMIC (LoRaWAN-in-C, [23] R. S. Sinha, Y. Wei and S. H. Hwang, “A survey on LPWA
formerly LoraMAC-in-C) framework provided by IBM”, technology: LoRa and NB-IoT”, Elsevier ICT Express, Vol. 3,
https://github.com/matthijskooijman/arduino-lmic, 2017. No. 1, pp. 14-21, 2017.
[4] Adafruit Inc., “Adafruit Feather 32u4 and M0 with HopeRF [24] M. Knight, “Reversing Lora”, https:://github.com/matt-
RFM95/96 LoRa Radio module, knight/research, 2016.
https://learn.adafruit.com/adafruit-feather-m0-radio-with-lora- [25] G. Kokkonis, S. Kontogiannis and D. Tomtsis, “A Smart IoT
radio-module/overview, 2016. Fuzzy Irrigation System”, IOSR Journal of Engineering, Vol.7,
No. 6, pp. 15-21, ISSN: 2250-3021, 2017.
[5] Microchip Inc., “RN2483 LoRa transceivers and modules
[26] A. Augustin, J. Yi, T. Clausen and W.M. Townsley, “A
datasheet and user’s manual”,
Study of LoRa: Long Range & Low Power Networks for the
http://www.microchip.com/wwwproducts/en/RN2483, 2010.
Internet of Things”, Sensors, Vol. 16, No. 9, 2016,
[6] Libelium company, “Libelium Waspmote networking guide”,
doi:10.3390/s16091466.
http://www.libelium.com/downloads/documentation/waspmote_lo
ra_868mhz_915mhz_sx1272_networking_guide.pdf, 2012.
[7] I. Arbel, S. Goldberg and A. Brosh, CattleSense Ltd,