Você está na página 1de 26

SMART ENVIRONMENT

MONITORING AND ANALYTIC IN


REAL-TIME SYSTEM
FOR VANNAMEI SHRIMP
Sritrusta Sukaridhoto, Dwi Kurnia Basuki, Jordan Maulana, Riyadh Arridha,
Rizqi Putri Nourma Budiarti and Tiyo Avianto

Politeknik Elektronika Negeri Surabaya, Politeknik Negeri Fakfak, Eyro Digital Teknologi
2

BACKGROUND

Most of Indonesian
shrimp farmer are:
still using traditional
method on shrimp
farming
they dont care about
water quality

ARIEET2017 Dec 12-14, 2017


3

BACKGROUND
Water Quality Parameters affect to Shrimp Farming [3]:
o Salinity
o Dissolved Oxygen
o Water temperature
o Potential of Hydrogen (pH)
o Turbidity

[3] Khairul Amri, and Iskandar Kanna. Budidaya Udang Vaname. N.p.: Gramedia Pustaka
Utama,
ARIEET2017
2013. Dec 12-14, 2017
4

SOLUTION:
SMART ENVIRONMENT
MONITORING AND
ANALYTICS IN REAL-
TIME SYSTEM (SEMAR)
FOR VANNAMEI
SHRIMP

ARIEET2017 Dec 12-14, 2017


5

CONTENTS
Background
Related Works
Internet of Things
Smart Environment Monitoring and Analytics in Real-time (SEMAR) System
System Design
Experiments
Conclusion and Future Works

ARIEET2017 Dec 12-14, 2017


6

INTERNET OF THINGS
IoT can be defined as a
set of interconnected
devices (humans, tags,
sensors, etc.) via the
Internet, which has the
ability to measure,
communicate and act
around the world

ARIEET2017 Dec 12-14, 2017


7

ARIEET2017 Dec 12-14, 2017


SMART ENVIRONMENT 8

MONITORING AND
ANALYTICS IN REAL-
TIME (SEMAR) SYSTEM[4]
Coral Reef Monitoring System [5]
Underwater Robot with Water
Quality Sensor[6]
Wireless Extension
Portable Low Cost Water Quality
Monitoring System[7]
Big Data Analytics[13]
Vannamei Shrimp Monitoring

ARIEET2017 Dec 12-14, 2017


9

SYSTEM DESIGN
BASED ON IOT REFERENCE MODEL
Collaboration & Process
Zeppelin, Mesosfer
Application
Spark MLLib, Web Visualization, Android Apps
Data Abstraction
Kafka Broker & MQTT Broker
Data Accumulation
PostgreSQL
Edge Computing
Raspberry PI, MQTT
Connectivity
4G
Physical device & Controller
Water Sensor, Raspberry PI, RC Boat

ARIEET2017 Dec 12-14, 2017


10

PHYSICAL DEVICE AND


CONTROLLER
subscribes it and Kaf
of MQTT to rece
tambakmonitoring,
data is tambakanalyt

F. Applications
The application la
1) Learning Process,
visualization in web
android notification.
(a)
G. Collaboration
Zeppelin is used
between analyst an
performed through w
Spark. Mesosfer [14]
Mesosfer is an IoT p
which enable user t
process.

IV

A. Water Quallity
Each parameter m
vannamei shrimp.
(b) Potential of
Figure 2. (a) Hardware Dimensions. (b) Hardware Placement Normal pH value f
is lethal for shrimp
B. Connectivity slowing down. Speci
ARIEET2017 Dec 12-14,
4G modem is used to connect Raspberry with Internet. The 2017
is between 7.5 and 8.
modem is powered by USB port from Raspberry PI and Dissolved O
11

CONNECTIVITY
4G Modem

ARIEET2017 Dec 12-14, 2017


12

EDGE COMPUTING
Software
Collect data from sensors
Send to server by MQTT Protocol (Pub-Sub architecture)
Server
MQTT Broker

ARIEET2017 Dec 12-14, 2017


13

DATA ACCUMULATION
PostgreSQL as DBMS

ARIEET2017 Dec 12-14, 2017


14

DATA ABSTRACTION
Real-time Visualization
Figure 15. An analytics system based on Big Data architecture.

5.2 Design and development of Integration Protocol for UUV system and 5D
World Map system
Internet of Things becomes popular in this research we designed a system to
monitor the environment by using sensors and transferred the data to Internet, analyzed
and visualize in real-time. In order to give real-time data, we developed Smart
Environment Monitoring and Analytics in Real-time (SEMAR) system. Figure 16 shows
about SEMAR architecture.
Data from sensors transferred to the Internet through several technologies such as
4G, or Wifi. In the sensor device, embedded system read the data from sensor and send the

Mobile Backend
data to the Internet by using MQTT protocol. In the server side, Mosquito server received
the data and deliver the data to Database and Analytic system through Kafka System. The
Database uses Big Data architecture. To visualize in the real-time we utilized node.js
technology.

29

MQTT Broker

Machine Learning

Sensor Sensor Sensor


ARIEET2017 Dec 12-14, 2017
15

APPLICATION

Machine Learning

Real-time Visualization
Figure 15. An analytics system based on Big Data architecture.

5.2 Design and development of Integration Protocol for UUV system and 5D
World Map system
Internet of Things becomes popular in this research we designed a system to
monitor the environment by using sensors and transferred the data to Internet, analyzed
Android Apps
and visualize in real-time. In order to give real-time data, we developed Smart
ARIEET2017
Environment Monitoring and Analytics in Real-time (SEMAR) system. Figure 16 shows Dec 12-14, 2017
water data from stressed in cause of high oxygen needing. Meanwhile, if
Spark Streaming temperature drops below 14oC, shrimp will be dead 16
Broker on topic
reSQL database. B. Rule Building
d to store water
The table consists
Table 1 show parameters value that influence shrimps life.
From the facts of water quality parameters given, we built set
RULES
, time. of rules to classify water quality for vannamei shrimp. It is
Water Quality Paramter
shown in table 2.

flow on big data Table 1. Parameter Value that influence in the shrimp's life
ata is received by Optimal
er to arrange data Params Bad Value Lethal Value
Value
connector Tempeerature
26-30 <25 & >30 <14
ribute data from (Celcius)
Potential of
Broker publishes Hydrogen (pH)
7.5-8.5 <5 >10
oducer stands for Dissolved
>3 <3 <1
application that Oxygen (ppm)

2180 1843 e-ISSN: 2289-8131 Vol. X No. X 3

ARIEET2017 Dec 12-14, 2017


17

RULES (2)
Supervised Machine Learning Classification Rule set

ARIEET2017 Dec 12-14, 2017


18

MACHINE LEARNING ALGORITHM

ARIEET2017 Dec 12-14, 2017


19

COLLABORATION AND FEEDBACK

Zeppelin Mesosfer
Mobile backend
Push Notification

ARIEET2017 Dec 12-14, 2017


Table 4. Linear SVM Dataset Training Linear Support Vector
Machine
20
0 0 5 0
True
1 0 1536 2

EXPERIMENT RESULTS
Label
2 0 0 143
Figure 3. ROC of D
0 1 2
Manuscript Title
Predicted Label
A. Confusion Matrix Results

Support Vector Machine Decision Tree


The SVM algorithm results some wrong prediction
Table 4 and 5 show the confusion matrix of training model
significantly. Every label has high number of wrong prediction
from Linear SVM and Decision Tree.
label.

Table 4. Linear SVM Dataset Training Linear Support Vector Table 5. Decision Tree Dataset Training
Machine
0 4 1 0
0 0 5 0
True
True 1 20 1536 2
1 0 1536 2 Label
Label
2 0 0 143
2 0 0 143
Figure 3. ROC of Dataset Using Linear Support Vector
0 1 2 Figure 4. ROC
0 1 2 Machine

Predicted Label Different from SVM, D


Predicted Label on all 3 classes that va
The SVM algorithm results some wrong prediction Different from SVM, Decision tree has higher accuracy of Decision Tree is used
significantly. Every label has high number of wrong prediction predicted label. So, with this comparement, we choosed
label. Decision Tree algorithm to be used for machine learning as it C. Real System Imp
has high accuracy of prediction. The experiment wa
ARIEET2017 Table 5. Decision Tree Dataset Training data with
Dec 12-14, 2017manually cla
B. Classification Results water quality data, then
Table 4 and 5 show the confusion matrix of training model
from Linear SVM and Decision Tree.
21
Table 4. Linear SVM Dataset Training Linear Support Vector
Machine

True
0 0 5 0 EXPERIMENT RESULTS
1 0 1536 2
Label
2 0 0 143
Figure 3. ROC of Dataset Using Linear Support Vector

SVM 0 1 Manuscript
2 Title Decision Tree Machine

Predicted Label
ining model The SVM algorithm results some wrong prediction
significantly. Every label has high number of wrong prediction
label.
port Vector
Table 5. Decision Tree Dataset Training

0 4 1 0
True
1 20 1536 2
Label
3 2 0 0 143
Figure 3. ROC of Dataset Using Linear Support Vector
Machine0 1 2 Figure 4. ROC of Dataset Using Decision Tree

ARIEET2017 Dec 12-14, 2017


Different from SVM, Decision Tree has much higher accuracy
Predicted Label
pr
22
fu
20
03

REAL EXPERIMENT D
D
Sm

Journal of Telecommunication, Electronic and Computer Engineering Experiment in 4 locations


Table 6. Comparison of Manual and Machine Learning Table 6 shows that at the time the experiment was conducted,
Classification in Real Water Pond
Comparison
the shrimp water pond is manual vs machine
at bad condition.
ACKNOWLEDGEMENT Then the system [1
Thelearning
immediately do its
authors earlylike
would warning feature,
to thanks to ER2Cpush notification
(EEPIS Robotics to
[2
android app. Center)
The activity is shown inProduction
Fig. 5 and Lab
Fig. for
6 shows
If
Research
bad classification
and Multimedia
send push the
the preparation
implemented andhardware
source code modification.
in water pond. This research is
notification
funded to smartphone
by KEMRISTEKDIKTI from Foreign Cooperation in
[3

2017 scheme with the number: 01/PL14/PG.1/SP2P/2017 and [4


03/PPKSK/ADDENDUM/INOVASI INDUSTERI-
DII/IX/2017 (Pengembangan Teknologi Ibeacon Sebagai
Device Dan Platform Internet Of Things (IOT) Big Data, Dan
Smart City).

REFERENCES [5
Table 6 shows that at the time the experiment was conducted,
the shrimp water pond is at bad condition. Then the system [1] Ekspor Udang Menurut Negara Tujuan Utama 2000-2015,
immediately do its early warning feature, push notification to https://www.bps.go.id/index.php/linkTabelStatis/1015, May 2017.
ARIEET2017 Dec
[2] Katherin Indriawati, Pembuatan modul kontrol kualitas air 12-14,udang
tambak 2017
android app. The activity is shown in Fig. 5 and Fig. 6 shows Figure 5. Early Warning Push notification to android app.
sebagai sarana pembelajaran perbaikan teknik budidaya udang, 2008.
Besari, Rahardhita Widyatra Sudibyo, and Sritrusta Sukaridhoto. "Design
and implementation of smart environment monitoring and analytics in
real-time system framework based on internet of underwater things and
23
big data." In Electronics Symposium (IES), 2016 International, pp. 403-
408. IEEE, 2016.
[5] Abdillah, Abid Famasya, Muhammad Herwindra Berlian, Yohanes

REAL EXPERIMENT
Yohanie Fridelin Panduman, Muhammad Aditya Wildan Akbar,
Marlanisa Arifatul Afifah, Anang Tjahjono, Sritrusta Sukaridhoto, and
Shiori Sasaki. "Design and Development of Low Cost Coral Monitoring
Figure 5. Early Warning Push notification to android app. System for Shallow Water Based on Internet of Underwater
Things." Journal of Telecommunication, Electronic and Computer
Engineering (JTEC) 9, no. 2-5 (2017): 97-101.
[6] Sukaridhoto, Sritrusta, Dadet Pramadihanto, Muhammad Alif, Andrie
Yuwono, and Nobuo Funabiki. "A design of radio-controlled submarine
modification for river water quality monitoring." In Intelligent
Technology and Its Applications (ISITIA), 2015 International Seminar
on, pp. 75-80. IEEE, 2015.
[7] Sukaridhoto, Sritrusta, Rahardhita Widyatra Sudibyo, Widi Sarinastiti,
Rizky Dharmawan, Atit Sasono, Ahmad Andika Saputra, and Shiori
Sasaki. "Design and development of a portable low-cost Sensors COTS-based
Calibration
water quality monitoring system." In Intelligent Technology and Its
Applications (ISITIA), 2016 International Seminar on, pp. 635-640.
IEEE, 2016.
[8] Rindang,Sensors
TegarCalibration
Esa. "Analisis dan pemetaan data parameter kualitas air
dengan menggunakan open rov.", 2016.
[9] Dissolved oxygen level monitoring in shrimp aquaculture using
embedded system. N.p.: Nectec, 23 Sept. 2010. PDF.
[10] Richter, Aaron N., Taghi M. Khoshgoftaar, Sara Landset, and Tawfiq
Figure 6. System implementation in shrimp water pond. Hasanin. "A multi-dimensional comparison of toolkits for machine
learning with big data." In Information Reuse and Integration (IRI), 2015
IEEE International Conference on, pp. 1-8. IEEE, 2015.
[11] Goib Wiranto, and I. Dewa Putu. Pembuatan sistem monitoring kualitas
VI. CONCLUSION AND FUTURE WORK air secara real time dan aplikasinya dalam pengelolaan tambak udang
Experiment
(2010).
[12] BaharuddinExperiment
Aziz, Elvayandri Muchtar, and Farkhad Ihsan Hariadi.
In this paper, we built shrimp monitoring system as an Human-Machine Interface for Water Quality Monitoring System of
extension of Smart Environment Monitoring and Analytics in
ARIEET2017 White-Legged Shrimp Pond (2017): n. pag. IEEE Explore Digital Library. Dec 12-14, 2017
Real-time System (SEMAR). Firstly, the water quality data is 27 Mar. 2017. Web. 14 Aug. 2017.
24

CONCLUSION & FUTURE WORKS


In this paper, we built shrimp monitoring system as an extension of Smart
Environment Monitoring and Analytics in Real-time System (SEMAR).
Firstly, the water quality data is fetched through water quality sensors. Then, it
is uploaded to the server by MQTT protocol. The server runs classification
based on training model.
Decision Tree method gives precise classification that its error is less than 1%
compare with SVM method that has 72% accuracy.
The result of data monitored and its classification can be seen on web and
android smartphone.
In future, this system is expected to be used as flexible as possible and
supports any aquaculture with only few configuration.

ARIEET2017 Dec 12-14, 2017


25

ACKNOWLEDGEMENT
The authors would like to thanks to ER2C (EEPIS Robotics Research Center)
and Multimedia Production Lab for the preparation and source code
modification.
This research is funded by
KEMRISTEKDIKTI from Foreign Cooperation in 2017 scheme with the number:
01/PL14/PG.1/SP2P/2017
03/PPKSK/ADDENDUM/INOV ASI INDUSTERI- DII/IX/2017 (Pengembangan
Teknologi Ibeacon Sebagai Device Dan Platform Internet Of Things (IOT) Big
Data, Dan Smart City).

ARIEET2017 Dec 12-14, 2017


ARIEET2017 Dec 12-14, 2017 26

THANK YOU
dhoto@pens.ac.id

Você também pode gostar