You are on page 1of 5

2015 IEEE Conference on Wireless Sensors

On the Application of IoT: Monitoring of Troughs


Water Level Using WSN
1
Lukas, 2Wisena Aditya Tanumihardja, 2Edy Gunawan,
1
Faculty of Engineering, Universitas Katolik Indonesia Atma Jaya, Jakarta, Indonesia
2,3
Red Diamond Laboratories, BSD City, Tangerang, Indonesia
1
lukas@atmajaya.ac.id, 2wisena.aditya@gmail.com, 3edy@potensi.net

Abstract—The need of monitoring the water level of troughs is This paper has the following structure: Section II reviews
increasing. This is parallel with the growing of Wireless Sensor the existing work on wireless sensor network in the farm. The
Network and Internet of Thing. By combining both approach, design and implementation of the system are discussed in
cattlemen can monitoring their troughs ubiquitous using their Section III. Section IV discusses the results of the experiments.
own personal device. This paper develop such system by using Finally, Section V concludes the paper.
LoRa™ as the media between sensor hub and nodes, while
Raspberry Pi is used as the gateway to push data into server. We
have successfully tested the functionality of node to hub join II. LITERATURE REVIEW
mechanism, where a node should join to the closest hub, which
marked by the highest RSSI of beacon signal. The designed In [1], wireless sensor network was used to help farmer
system is able to wirelessly sense trough water level up to 50 monitor their surrounding environment such as humidity,
simulated nodes. temperature, and also movement of animals. These data are
transmitted to the central hub using wireless ZigBee that had
Keywords—Wireless Sensor Network; Internet of Thing; center frequency 2.4GHz. For the main controller they used
LoRa™; RFM95; ATMEGA; Raspberry Pi Microchip PIC24F16KA102 that capable to read analog data
that produced by the sensors. This system was also claimed to
I. INTRODUCTION reach 1.5 km maximum at 2.4GHz ISM band.
In the era of wireless technology, wireless sensor network In [2], nRF24L01 was used to transmit the axis data of a
(WSN) has become an alternative to reach a remote area or an chicken movement and the body temperature for avian
area that require no cable in it. Compared to the wiring method, influenza surveillance. The device itself was attached on the
wireless is cheaper, since it is free of cable maintenance, chicken, thus it should be battery operated. Along with
flexible in installation, and also reliable. Many WSN microcontroller C8051F411 that operated with button battery,
applications have been developed such in the farm, to monitor which may operate up to 2 year monitoring without
the environment conditions [1], even an early detection of
replacement.
avian influenza in poultry farm [2]. Along with the wireless
era, the Internet of Things (IoT) has rapidly growing with its Paper [4] discussed the Long Range (LoRa™) technology
principal to bring any devices or things are ubiquitous that was developed by semiconductor manufacturer, Semtech.
connected with any desktops or portable devices [3]. It is shown that the spread spectrum modulation that utilized
As an application among the others, that combine the WSN broadband linear frequency modulated in LoRa™, can reach
and IoT, our group has developed a system to help the up to 6.6 km with Packet Error Rate 7.3% at 838MHz of
livestock productivity by monitoring the water level in a center frequency.
trough. Although there are troughs that have auto refill system,
III. METHODOLOGY
but monitoring the quantity of the water in troughs have not
been implemented. In some cases, the auto refill were failed to
A. Topology
fill the trough and caused water underflow, or others were
failed to stop refill the trough which caused water overflow. The basic idea of IoT is every node connected to the
internet. However, in our system, the nodes is not directly
Designing trough monitoring system has several aspects connected to the internet, or could be explained that the nodes
that need to be considered. Most of the troughs are placed should use a gateway to connect to the internet. The reason
inside the barn which not provided with electricity. Therefore, behind this design, is based on the low power consumption
the system has to be battery operated. On the other hand, high requirement, since each of the node will be placed on the field,
transmission power is needed since the distance between where no electricity can be plugged, thus it requires battery
central hub and the barn has more than 2 km. Further, when operated. As for the gateway itself, it will be placed up on the
there are several barns and each barn has many troughs, the air roof of cattlemans’ house, where it can use the power line to
traffic communication will be crowded, causing the data operate.
transaction to be failed.
Since the low power consumption of each node is
emphasized in this system, thus we chose the star topology as a

978-1-4673-9398-0/15/$31.00 ©2015 IEEE 58


2015 IEEE Conference on Wireless Sensors

default communication topology between the hub and nodes. In Since the target user of this system are the cattleman who is
this paper, the hub refer to sensor hub, where it collects all expected to have minimum engineering background, thus the
sensors’ data from nodes, and a node is refer to the device that whole system needs to be automatically self-configured. At the
is placed at the trough to sense the water level inside the beginning, the sensor hub will broadcast a beacon to the air.
trough. The data from nodes, will then be send to the server Meanwhile, the nodes will be scanning for surrounding
trough gateway, where in the next section we will explain more beacons. As the node detect and find the highest Received
about the design of the gateway, sensor hub, and nodes further. Signal Strength Indicator (RSSI) from the node side, it will join
to that beacon. Once the node sends a request join packet to the
B. Design desired hub, it then will receive a reply of its logical address
On the field, the nearest distance between cattleman’s from the hub. In unresponsive request from the node with
house and the barns is around 1 km and the furthest is 3 km. duration over 30 minutes, the hub will mark the node with non-
There is also consideration that the elevation difference active status.
between hub and nodes will affect the transmission, where
most of the nodes, will be placed near to the ground due to the
TABLE I. TROUGH CONDITION REPRESENTED BY THE STATUS OF FLOAT
height of the trough is less than 50 cm, where the hub’s SWITCHES
antenna will be placed above the cattleman’s house which
could be 8 meter high from the ground. With regard to this Float Switch Upper Float Switch Lower Condition
requirement and also due to the free frequency that can be use
on the application’s country, our preliminary design of the 1 1 Full
system is to use the LoRa™ with 915 MHz center frequency.
The alternative product of LoRa™, HopeRF RFM95 [5] is 0 1 Normal
used in this system. As the controller in the node side we use
0 0 Empty
Atmel ATmega328 [6] to read the water condition in the trough
which we use the float switch GE-1307 [7]. As for the sensor
hub, we use Raspberry Pi 1 model B with 512MB of random Then, within an hour of nonactivity, the hub will discard the
access memory (RAM). ATmega is chosen for the nodes to node from its list. To resume the connection, the node should
satisfy the low power system for remote area purpose, while repeat the request join procedure again.
the choice of Raspberry Pi as the controller for the hub is based
on the system on chip and complete embedded system that has In a regular condition, all the nodes will send periodic data
SPI to communicate with HopeRF RFM95. which inform the water level status of the troughs, where it is
categorized into three conditions: underflow, normal, and
overflow. This category is designed to satisfy the number of
data that has to be transmitted which influence the power
Server consumption per transmission. As a limitation, in this paper,
the data confidentiality has not been considered yet. However,
the data integrity check by using CRC method is on the list of
TCP/IP our system’s feature, since it has been implemented in the
packet structure for LoRa™.
Gateway

Sensor Hub Gateway


Node Controller
(Raspberry Pi
(ATMEGA 328)
LoRaTM v.1 type B)
915 MHz
SPI SPI
Node Node Node
1 2 n Sensor Hub Node
Transmitter Transmitter
(HopeRF RFM95) (HopeRF RFM95)
Fig. 1. Topology of trough water level monitoring system
(a) (b)

To display the status of troughs it uses HDMI port, while the Fig. 2. Connection diagram of trough water level monitoring
3.5mm jacks can be used as output to an alarm system through system, (a) Hub and (b) Node
active speaker, and ethernet port as a gateway physical device.
The overall system diagram can be seen in the Fig. 2. For the
sensor itself there are two float switches which placed as
shown in the Fig. 3. These two switches will mark three
conditions of the water level in trough as shown in Table I.

59
2015 IEEE Conference on Wireless Sensors

IV. RESULT AND DISCUSSION We were also looking for the behavior of the system, that have
caused the nodes failed to join or sending the data. The
A. Functional test in the Laboratory experiment showed that, by increasing the number of nodes, it
As described in the Sec. I, a float switch will be placed will also increase the collision inside the whole monitoring
inside each trough to measure the water level in the trough, systems. In our experiments we simulated 100 nodes present in
which is still simulated in our experiments. In the first test, we the system using 5 physical nodes, by increasing the number of
gather a node and two sensor hubs on a table with the data sending per period. In the normal condition, the node will
following order: first hub --- second hub --- sensor node, which sleep for 10 minutes before it sends the next data.
is illustrated in Fig. 4. We then powered on the hubs and node. In the condition where heavy traffic occurred in the
Once the hub is active, it then spreads a beacon and node will transmission line, the hub would not be able to give an
keep listening other beacons for over 1 minute. This 1 minute acknowledgement to the node either node's join request or
period was sufficient in the lab condition but it still needs to be node's periodic data. Although each node is able to listen to the
adjusted in the field. From the node log data that were captured, air traffic being used or not, but two nodes would have been
it showed that 2 beacons were appeared. The first beacon had waiting and finally decide to send data at the same time when
RSSI of -24dBm and marked with unique ID “Hub 1”. The they saw that air traffic is none. In Table III the result of packet
second beacon had RSSI of -17 dBm and marked with a error with simulated nodes is presented, along with several
numbers of nodes. In the experiment we used 26 bytes of
payload length, and each node was set to send 100 packets with
sending interval n second, where n was set by dividing the
normal sleep time with number of nodes. The experiment was
Float Switch Upper repeated 50 times and shown in the Table III as the mean of
valid packet.

TABLE II. FLOAT SWITCHES LAB TEST RESULTS WITH DISTANCE 50 CM

Float Switch Bottom Float Switch Float Switch Data at Sensor


Data at Sensor Hub
Upper Lower Node

1 1 OF OF

0 1 NE NE
Trough
0 0 E E

Fig. 3. Placement diagram of float switches inside a trough 1 0 ERR ERR

unique ID “Hub2”. This RSSI results indicate that the longer TABLE III. RESULT OF COMMUNICATION ERROR IN HUB SIDE CAUSE BY
distance will increase the loss of signal power during HEAVY TRAFFIC
transmission.
Number of Periodic Data Number of Packet Valid 26 bytes
As the node trying to connect to one of the sensor hubs, it Nodes Interval (s) Received Packet
showed that Hub 2 was chosen since it had a bigger RSSI. We
5 600 100 100
then manually controlled the float switches to simulate the
possible conditions. The result is shown by the Table II, where 10 300 100 100
we also compared the data that was read by the node from the 25 120 100 99
float switch and the data that transmitted from node to the hub. 50 60 100 100
The notation of “OF” stands for “overflow” status, “NE” stands
for “not empty”, “E” stands for “empty”, and “ERR” stands for 75 40 100 91
“error”. The error condition might not be appear in the real 100 30 100 83
condition except there was a hardware problem, which is why
we named it as “ERR”.
B. Field Test
As we mentioned in the Sec. III, that most of the nodes
Hub Hub will be placed close to the ground with approximated height is
Node less than 50 cm. Therefore in this test, we analyze the behavior
1 2
of the transmission between hub and node that has different
50 cm 50 cm elevation. In the test, we also analyze the distance between
hub and node as a comparison to the vertical position. On the
Fig. 4. Functionality test between one sensor node and two sensor hubs other hand, to maximize the transmission, we analyze the
antenna polarization difference between the horizontal and
vertical.

60
2015 IEEE Conference on Wireless Sensors

TABLE IV. THE RSSI OF NODES’ LINK WITH ELEVATION AND ANTENNA POLARIZATION DIFFERENCE

Hub Elevation (m) Node Elevation (m) Distance (km) Polarization RSSI (dbm)
Horizontal -101
0.5
Vertical -103
Horizontal -103
1
Vertical Lost
Horizontal -103
0.4 1.5
Vertical Lost
Horizontal -103
2.2
Vertical Lost
Horzontal Lost
2.7
Vertical Lost
2.5
Horizontal -97
0.5
Vertical -98
Horizontal -101
1
Vertical -100
Horizontal -101
6 1.5
Vertical -102
Horizontal -96
2.2
Vertical -98
Horizontal -103
2.7
Vertical -103
Horizontal Lost
0.5
Vertical Lost
Horizontal Lost
1
Vertical Lost
Horizontal Lost
0.4 1.5
Vertical Lost
Horizontal Lost
2.2
Vertical Lost
Horizontal Lost
2.7
Vertical Lost
8
Horizontal -96
0.5
Vertical -98
Horizontal -100
1
Vertical -101
Horizontal -100
6 1.5
Vertical -102
Horizontal -103
2.2
Vertical -103
Horizontal -103
2.7
Vertical -104

61
2015 IEEE Conference on Wireless Sensors

As the results, which are shown in Table IV, the smaller During the laboratory test, the basic mechanism has been
difference between hub and node elevation will affect the successfully implemented that marked with successful
succcessful rate of transmission, which in our case indicated transaction between hub and nodes, which include node joining
by the RSSI, with dbm as the unit. These results, confirm our to hub and periodic sensor data by node. However, some
hypothesis about the transmission behavior if the elevation mechanisms need to be improved in this system such as the air
difference between hub and node is large. It indicates in our traffic when a hub has to handle more than a hundred of nodes.
results as Lost in packet transmission. The best result that we Based on our field test, the bigger elevation difference between
can achieved, when the node is 6m high above the ground and hub and node, could result in a bad transmission quality.
However, this could be overcome by lowering the hubs’
the hub is place 8m high above the ground. Also based on our
antenna or higher-up the nodes’ antenna which reduce the
test results, although when the node is place 0.4m above the
elevation difference between hub and node. Field experiments
ground and the hub 2.5m above the ground, the result is also showed that, it will be best to use horizontal antenna
beyond with the node that place 6 meter above the ground. We polarization for our application.
conclude that the closer our device is placed to the ground,
will affect the transmission quality, which tend to be poor. On
REFERENCES
the other hand, based on field experiment’s results which also
shown in Table IV, the antenna polarization is also affect the [1] Nor A.K, Asral.B.B.J, Liew J.H., and Uda H., “Wireless Sensor Node
for Farm Monitoring”, Circuit and System (ICCAS), 2013,
link quality in the close distance between hub and node (we International Conference on, pp. 150-153, 18-19 Sept 2013
refer the close distance as hub and node difference ≤ 1.5m). [2] Hironao O., Koutarou S., Tsukamoto K., and Toshiro I., “Avian
Finally, in our analysis, the best antenna polarization for our influenza surveillance system in poultry farms using wireless sensor
application is the horizontal one. But the effect of antenna network”, Design Test Integration and Packaging of MEMS/MOEMS
(DTIP), 2010, Symposium on, pp. 253-258 5-7 May 2010.
polarization might be different in other environment. [3] Lu Tan and Neng Wang, “Future Internet: The Internet of Things”,
Advance Computer Theory and Engineering, 2010, 3rd International
V. CONCLUSION Conference on, vol. 5, pp. 376-380, 20-22 Aug 2010.
The wireless sensor network provides a great solution to [4] Mohamed A. and Axel S.,“Free Space Range Measurements with
Semtech LoRa TM Technology”, Wireless System within the
minimize the wired communication which costs a lot when two Conference on Intelligent Data Acquisition and Advance Computer
devices are located far apart. On the other hand, IoT is a Systems, 2014, 2nd IEEE International Symposium on, pp 19.23, 11-12
promising solution to communicate between devices or Sept 2014.
between device and user not only in local area but in the wider [5] HopeRF, “RFM95W Low Power Long Range Transceiver Module”,
area that has internet coverage. In this paper, we have applied http://www.hoperf.com/rf/lora/RFM95W.htm
[6] Atmel, “ATmega328”, http://www.atmel.com/devices/atmega328.aspx
wireless sensor network and IoT to monitoring water level of [7] A. Yite, “GE-1307 Plastic Level Switch|Float Switch”Float switch
troughs, thus cattleman can remotely observe their livestock. http://www.ayite.net/page/1106/GE-1307.htm

62

You might also like