See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330273027

An Automated Irrigation System for Smart Agriculture Using the Internet of

Things

Conference Paper · November 2018

DOI: 10.1109/ICARCV.2018.8581221

CITATIONS READS
0 233

3 authors, including:

Ramalakshmi Ramar Seshadhri Srinivasan

Kalasalingam Academy of Research and Education (Deemed to be University) ABB
14 PUBLICATIONS   35 CITATIONS    120 PUBLICATIONS   400 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Self-Calibration of Dead Reckoning Sensor for Skid-Steer Mobile Robot Localization Using Neuro-Fuzzy Systems View project

Nonlinear control: theory, symbolic software tools and applications View project

All content following this page was uploaded by Ramalakshmi Ramar on 11 January 2019.

The user has requested enhancement of the downloaded file.

2018 15th International Conference on
Control, Automation, Robotics and Vision (ICARCV)
Singapore, November 18-21, 2018

An Automated Irrigation System for Smart Agriculture Using the

Internet of Things
V. Ramachandran, R. Ramalakshmi, and Seshadhri Srinivasan

Abstract— Water is a vital and scarce resource in agriculture agricultural environmental information was studied in [10].
and its optimal management is emerging as a key challenge. However, control functionalities were not studied in these
This paper presents an automated irrigation system to reduce investigation. The investigations in [11] and [12] proposed
water utilization in agriculture by combining the Internet of
Things (IoT), cloud computing and optimization tools. The a framework with limited control functions. The role of IoT
automated irrigation system deploys low cost sensors to sense for controlling the water consumption in irrigation has been
variables of interest such as soil moisture, pH, soil type, and studied by many scientists as well. A simple irrigation system
weather conditions. The data is stored in Thingspeak cloud was studied in [13]. An advanced system was proposed
service for monitoring and data-storage. The field data is in [14] that aims to transform traditional farming to modern
transmitted to the cloud using Wi-Fi modem and using GSM
cellular networks. Then an optimization model is used to one. The investigation in [15] proposed using wireless sensor
compute the optimal irrigation rate which are automated using networks for managing irrigation in agricultural farms. The
a solenoid valve controlled using an ARM controller (WEMOS framework proposed allowed user to interact with the data
D1). The variables of interest are stored in the cloud and and consult in a comfortable way. An IoT based smart farm
offered as a service to the farmers. The proposed approach irrigation system was proposed in [16] wherein Zigbee was
is demonstrated on a pilot scale agricultural facility and our
results demonstrate the reduction in water utilization, increase used for communication between sensor nodes and base
in data-availability, and visualization. station. While data was collected and processed in these
investigation, combining cloud computing with IoT was not
I. I NTRODUCTION fully explored. This provides additional opportunities due to
Sustaining agricultural productivity, guaranteeing food- the service delivery models of the cloud.
security, and enhancing economic growth in the face of Recently significant efforts have been devoted to combine
climate variability, diminishing labour force, and changing IoT and cloud computing [17], [19], [20] showed that the IoT
soil conditions requires innovation in agriculture. In India, benefits can be enhanced by combining it with cloud comput-
agriculture contributes 18% of the country’s Gross Domestic ing. The role of cloud-based IoT scheme for precision agri-
Product (GDP) and employs more than 50% of the popu- culture was studied in [18]. However, the final control aspect
lation [1]. Notwithstanding these contributions, the sector is has not been considered. Similarly, optimization models for
under stress and the recent economic survey of the Indian irrigation have been studied in [21] without discussions on
government has pointed out the need to extract “more crop monitoring and control. A review of the literature reveals that
per drop” which indicates exploiting technology and good existing approaches on using IoT are restricted to monitoring
practices to enhance productivity per drop of water. While and data-aggregation. Final control including optimization
this is largely dependent on the irrigation system, recent of resources has not been studied in the literature to our
developments in technology are being touted as solutions [2]. best knowledge. In this paper, we combine the IoT sensing
The real-time environmental parameters such as temperature, and networking capability with cloud interfaces, use the data
soil moisture, humidity, evapotranspiration, cropping cycles, to study the optimal irrigation rates, and finally implement
and others influence the crop lifecycle. Resource utilization the computed flow rates by commanding a solenoid valve.
can be optimized by real-time monitoring of these parameters Consequently, a comprehensive solution including sensing,
and taking corrective actions based on sensed information. networking, control, and optimization is proposed. Such a
In this backdrop, the Internet of Things (IoT) with has methodology has not been proposed for agricultural irrigation
emerged as an enabler of agriculture automation [3], [4], purposes to our best knowledge. The main contributions of
[5]. The IoT uses recent advances in sensing, networking, the paper are:
and computing technologies to enable novel applications
and services. The use of IoT for crop monitoring has been 1) A smart irrigation system that uses IoT and cloud-
studied in [6]-[9]. Fusing crop statistical information and connectivity to aggregate and store information, an
optimization model to compute the optimal irriga-
This research is funded by the DST through the Initiative to Promote tion parameters, and final control implemented using
Habitat Energy Efficiency Project title: Energy Savings Through IoT and
Nonlinear Model Predictive Controller (TMD/CERI/BEE/2016/088) solenoid valves.
V. Ramachandran and R. Ramalakshmi are with the Dept. of Computer 2) Design aspects of IoT hardware, software, and their
Science and Engineering, Kalasalingam Academy of Research and Educa- integration along with networking as well as cloud
tion, India 626126 e-mail: (v.ramachndran@klu.ac.in,rama@ klu.ac.in)
Seshadhri Srinivasan is with Berkeley Education Alliance for Research connectivity are discussed.
in Singapore e-mail: (seshadhri@ieee.org) 3) Demonstrate the control methodology and hardware

978-1-5386-9582-1/18/$31.00 ©2018 IEEE 210

 Device Specifications
using experiments/simulations. WEMOS D1 Controller ESP-8266EX
The rest of the paper is organized as follows. Section II Soil Moisture Sensor YL69
Solenoid Valve 1/2 Inch, 12V
presents the system architecture and design which describes Gardening sprinkler 4-hole female
the various components of the system. The optimization Drip irrigation hose 4 mm, 2 meter (length)
model for irrigation control is presented in Section III. Water Flow Sensor YF-201
Humidity & Temperature sensor DHT 11
Results and observations are presented in Section IV. Con- pH sensor pH meter probe (0-10)
clusions are drawn from the obtained results and analysis in
Section V. TABLE I: Components of the Architecture

II. S YSTEM A RCHITECTURE AND D ESIGN

Fig. 2: Organization of Agriculture Land as Sectors

Fig. 1: IoT Architecture of the Smart Irrigation System

described the architecture, we provide a succinct description
The IoT architecture used for implementing the precision of the different components used in the hardware.
irrigation system is shown in Fig. 1. The components of 1) WEMOS D1 Controller: The D1 is an ESP8266 (Wi-
the architecture are listed in Tab. I. The components are Fi) based controller which is compatible with the Arduino
selected based on cost-reliability analysis. While the low-cost IDE. The functions are same as the Arduino Uno controller,
sensors lead to reliability issues, they perform reasonably whereas the WEMOS has the ESP8266 module by default
well for the application considered. In this work low cost on the board hence reducing the complexity of interfacing
soil moisture and flow sensors are used. The sensors send the an ESP8266 with Arduino Uno. ADCs(Analog to Digital
data to the WEMOS D1 controller and the controller controls convertors) were used to interface multiple analog sensors
the flow using sensed information. The controller controls to the controller.
the solenoid valve through which the field is irrigated. The 2) Solenoid Valve: The traditional valves are replaced
controller also controls the DC motor on/off state since the with solenoid valves to control the flow. The valve is oper-
motor has to be on only when any one of the valve is ated either in ON/OFF mode and a pulse-width modulation
in open state. The controller is connected to the internet approach is used to control the flow, i.e., the amount of flow
through the GSM GPRS module. The internet connectivity is proportional to the time-period for which the valve remains
is provided through GSM as broadband is not feasible in in ON state over a given time period. A 24 V relay is used
many rural agricultural areas, whereas more than 70% of the to turn ON/OFF the solenoid valve.
land in India is feasible with GSM based cellular network. 3) Soil Moisture Sensor: In our design, YL69 series soil
The information retrieved from the field is used to control moisture sensor or probe measures the volumetric water
the irrigation system, and it is also stored in the cloud content in the soil. Determining soil moisture is considered
(Thingspeak) for further analysis. Remote monitoring of as an important task in agriculture to assist farmers manage
the field was provided through web interface and mobile the irrigation systems more effectively. Compared to other
interface. The farm is split into several sectors as depicted low cost sensor such as gypsum block sensors, these probes
in Fig. 2, the water flows through different valves for tender a rapid response time. Due to this reason the sensor
each sector which are deployed with a set of sensors for is chosen and used in the proposed design. Placing the soil
monitoring and a solenoid valve as an actuator. moisture sensor in the right place is very important, since
The sectors are sort of control regimes for which the a sector’s irrigation is controlled by the value of the soil
water can be irrigated. This helps organizing the irrigation moisture sensor deployed for that sector. The soil moisture
and monitoring to meet the needs of the individual sec- sensor works between 3.3V and 5V power supply. The output
tors, thereby better management can be achieved. Having value of the sensor is between 0 ohms to 1000 ohms. Based

211
 Soil Type Resistance value range Time stamp Entry Id Soil Moisture Flow in
Dry 0- 300 Ω `
in ω s
Humid 300-700 Ω 2018-03-26 10:38:51 UTC 94 310.00 6.12
Wet 700-100 Ω 2018-03-26 10:39:12 UTC 95 316.00 6.21
2018-03-26 10:39:33 UTC 96 330.00 6.29
TABLE II: Resistance range of Soil Moisture Sensor 2018-03-26 10:39:54 UTC 97 345.00 6.39
2018-03-26 10:40:15 UTC 98 368.00 6.51
2018-03-26 10:40:36 UTC 99 382.00 6.62

on the sensor reading the soil can be classified into Dry, TABLE III: Sample Data Stored in Thingspeak
Humid and wet. The soil moisture sensor used is connected
to the analog pin of the controller through wire. The sensor
A. Cloud-based Remote Monitoring
range for determining the soil type is shown in Tab. II.
The Irrigation system is initiated based on the soil moisture
4) Flow Sensor: In this system flow sensor (YF-201) is
sensor value. The field is irrigated automatically using the
used to measure the amount of water utilized in the process of
solenoid valve or sprinkler. The solenoid valve and the
irrigation. The amount of water utilized has to be measured
sprinkler were connected to the controller using relay switch.
in the experimental setup for the traditional and automated
The data from the controller is transmitted to the Thingspeak
irrigation methods so that comparison can be easily carried
cloud, and the data can be viewed using Thingspeak website.
out. The water flow sensor is aligned in parallel with the
The user interface is a simple monitoring interface which
water line, and a pinwheel inside the sensor is used for
shows the readings from the sensor, the solenoid valve status
measuring the water irrigated through it. The water flow is
and the amount of water used. The screen shot of the
measured in litres/second.
ThingSpeak web interface is shown in Fig. 4 and the snippet
5) Data Transmission: Transferring the collected infor- of the data displayed is shown in Table III.
mation from the farm to the Internet is a major issue as Similarly the PH value of the soil, Humidity and temper-
internet connectivity through broadband to agricultural area ature were also stored in cloud. Thingspeak cloud service is
is still an infeasible solution in more than 50% of the an easy to access service and has inbuilt lab view functions.
agricultural lands across India. In this work the data is Fig. 4 shows the Thingspeak Cloud Service Interface for a
transmitted using the GPRS internet connectivity available smart phone wherein the data are represented in charts. The
through the cellular network providers. The cellular network data feeds can be stored in the Thingspeak along with the
covers majority of the agricultural lands in India and with timestamp for further analysis and for providing real-time
the emergence of 3G and 4G technology it is possible to data visualization.
transmit data quickly. Instead of using a GPRS module along
with the controller, we have used a Wi-Fi hotspot device for
data transmission from the controller to the internet as the
transmission speed is high when using hot spot as compared
with a GSM GPRS module of the controller. The use of
WEMOS controller has reduced the complexity in interfacing
an ESP8266 when compared to Arduino. The GSM module
is used to send messages from the controller regarding the
status of the field.

Fig. 4: ThingSpeak Interface for Smart Phone

III. O PTIMIZATION M ODEL

Computing the minimum irrigation rates based on sensed
information is a decision making problem requiring help
of optimization tools. Therefore, we model the optimal
irrigation problem as an optimization problem. We define
the following:
Fig. 3: ThingSpeak Web Interface
Definition 1: The difference between the planting and har-
vesting time is denoted as the irrigation period.

212
Definition 2: Maximal rainfall over the irrigation period by:
denoted by Rmax is the upper bound on the rainfall during
M:
the period.
min w1 (t)St + w2 (t)rt + w3 (t)Q(t)∆
Our objective is to minimize the irrigation ofPthe water Q(t)
over a 24 hour horizon which is given by I = t = Q∆t s.t.
where Q denotes the flow and ∆ unit time step used in the
Rmin ≤ rt ≤ Rmax ,
analysis. In addition, we aim to exploit the use of rain water,
t+T
X∆
and soil moisture content. The objective is modelled as:
Q(t)∆ ≤ I
t=t+∆

J = w1 (t)St + w2 rt + w3 (t)Q(t)∆ (1) (et − rt − St + dt )

Q(t) ≥
∀t ∈ {t + ∆, t + 2∆, . . . , t + T ∆} ∆
W Rmax (t) − rt − St + dt
Q(t) ≤
∆
where w1 , w2 and w3 are weighing factors, St soil moisture EP (t) ≤ rt + St + Qi (t)∆ − dt
content at time t, and the weights are selected depending on Qmin ≤ Q(t) ≤ Qmax
the crop or soil conditions. rt , St , dt , Q(t) ≥ 0,
Limits on the rainfall r are given by w1 + w2 + w3 = 1
∀t ∈ {t + ∆, t + 2∆, . . . , t + T ∆}
Rmin ≤ rt ≤ Rmax ∀t ∈ {t + ∆, t + 2∆, . . . , t + T ∆} The optimization model M is a linear programming problem
(2) and can be solved with open source solvers such as Gnu
Linear Programming Kit on single-board computers such as
BeagleBone Black. However, in our analysis, the problem
where Rmax and Rmin denote the maximum and minimum
was solved in a computer using MATLAB’s linprog routine.
values of the rainfall during the period. The upper limit on
the irrigation is given by IV. R ESULTS
A. Real-time Experiments
t+T
X∆ In our experiments, a pilot having four land sectors each
Q(t)∆ ≤ I (3) 2 × 2 square feet were taken to test the method. One
t=t+∆
sector was irrigated using traditional method in which the
water flow was controlled manually, and the other three
Following [21], the irrigation at time periods is limited by (Automated Irrigation, Drip Irrigation, Sprinkler Irrigation)
were irrigated with the automated method using sensors and
actuators (solenoid valve, sprinkler, and Drip). The spinach
(et − rt − St + dt ) named Amaranthus tricolor seeds was sown in even quantity
Q(t) ≥ (4)
∆ and grown in the entire four land sector. Water flow sensor,
W Rmax (t) − rt − St + dt soil moisture sensor, soil PH sensor, was installed in all the
Q(t) ≤
∆ sectors. Weather sensors like humidity, temperature Sensor
and rain sensor were deployed in common for all the four
where et is the threshold on water use, dt water drained, and sectors. The actuators were connected to the controller using
W Rmax maximum water reserve, respectively. In addition, wired connection. The input to the sprinkler was given using
the evapotranspiration rate is constrained by a low pressure water pipe as the land sector taken for
irrigation is small. The irrigation was done based on the
moisture sensor value. Different moisture values are set for
EP (t) ≤ rt + St + Q(t)∆ − dt ∀t (5) different crops, in the experimental set up the value was
set to less than 300 ohms to 950 ohms, if the value drops
below 300 ohms then the solenoid valves were opened and
In addition, the rainfall, soil moisture, flow-rate, and drain the field is irrigated and if the value was greater than 950
are all positive real-values and this is expressed as ohms then the solenoid valves were closed. For drip irrigation
automation, the solenoid valves were used to supply water
to the drip irrigation tube. In the sprinkler irrigation system
rt , St , dt , Q(t) ≥ 0 (6) sprinkler is supplied with a low pressure water input as the
experimental prototype is for demonstration purpose. The
soil moisture metric used to automate the irrigation makes
The optimization model for reducing the irrigation is given sure that the land is not dry at any point of time. The data

213
is uploaded to the things speak cloud using write API key. 15
The experiments were conducted for a period of 3 weeks. To Automated
Manual

Irrigation in ltrs/day
test the effectiveness of the optimization approach, we first
propose a moisture based control as shown in Fig. 5. 10

0
1 2 3 4 5 6
days
Fig. 6: Automated versus Manual Control with sprinkler
system for flow-based control

method and an increase in 7% savings over flow-based con-

trol. Similarly, in sprinkler irrigation, the savings were 26%
and 22% in solenoid valve based control. Our simulations
shows that optimization based control outperforms the flow-
based control in terms of water savings.

15
Fig. 5: Moisture-based Control Automated
Manual
Irrigation in ltrs/day

The following two scenarios were compared:

• Flow control based on moisture level with valve control,
10
sprinkler control, and drip irrigation;
• Flow control using optimization approach with valve
irrigation, sprinkler control, and drip irrigation; 5
In the optimization model, the change in irrigation conditions
correlate to change in the limits of the flow-rates Q(t).

B. Results with Heuristic Control 0

1 2 3 4 5 6
The flows for a period of 3 weeks were used to study days
the effectiveness of the flow-based control. The results of
all three automated irrigation methods were compared with Fig. 7: Automated versus Manual Control with Sprinkler
conventional method. In Drip irrigation method the system System for optimization-based control
is highly efficient saving around 24%, compared to 20%
in sprinkler and 16% in Solenoid valve based automated
irrigation. In this experimental prototype, deficient watering D. Observations
condition was eliminated as the water resources were suffi- • It was observed that the pH value of the soil decreases
cient throughout the experiment. Water deficit might occur with an increase in moisture levels.
when the system is implemented in the real agricultural field • The IoT and cloud-connectivity enhanced the data-
due to water scarcity. The data is stored in cloud with ease aggregation and visualization capability significantly.
using the API funtions in thingspeak. The flow control over • Combining IoT, cloud-connectivity, and optimization
for a period of six days with flow-based control is shown models will help enhance water efficiency of the agri-
in Fig. 6 shows that the drip irrigation the flow is relatively culture systems.
lower than other irrigation systems. • The irrigation system was automated by connecting
solenoid values which helped increase the agility of the
C. Simulation Results with Optimization Based Control control.
The flows for a period of 1 week with optimization based
control was used to study the effectiveness of optimization V. C ONCLUSIONS
based control for the three irrigation schemes. In drip irriga- This investigation presented an automated irrigation sys-
tion method the method provided 31.2% over conventional tem to reduce water utilization in agriculture by combining

214
 Internet of Things (IoT), cloud computing, and optimization. [15] Navarro-Hellı́n, H., Torres-Sánchez, R., Soto-Valles, F., Albaladejo-
The automated irrigation system is realized by deploying Pérez, C., López-Riquelme, J. A., & Domingo-Miguel, R. (2015). A
wireless sensors architecture for efficient irrigation water management.
low-cost sensors to sense variables of interest such as pH, Agricultural Water Management, 151, 64-74.
temperature, humidity, soil type and weather conditions. The [16] Saraf, Shweta B., and Dhanashri H. Gawali. ”IoT based smart irriga-
data is stored in Thingspeak cloud service for monitoring tion monitoring and controlling system.” In Recent Trends in Electron-
ics, Information & Communication Technology (RTEICT), 2017 2nd
and storage. Then an optimization model for reducing the IEEE International Conference on, pp. 815-819. IEEE, 2017.
water usage was proposed and constraints modelling the [17] TongKe, F. (2013). Smart agriculture based on cloud computing and
physical conditions were included. The optimal flow rate was IOT. Journal of Convergence Information Technology, 8(2).
[18] Khattab, Ahmed, Ahmed Abdelgawad, and Kumar Yelmarthi. ”Design
determined solving the optimization model and it was shown and implementation of a cloud-based IoT scheme for precision agricul-
that the flow rate can be automated using solenoid valves. ture.” Microelectronics (ICM), 2016 28th International Conference on.
The optimization-based control was compared with flow- IEEE, 2016.
[19] Patil, V. C., Al-Gaadi, K. A., Biradar, D. P., & Rangaswamy, M.
based control and our results demonstrated that optimization (2012). Internet of things (Iot) and cloud computing for agriculture: An
models help in reducing the water consumption. Improving overview. Proceedings of Agro-Informatics and Precision Agriculture
the optimization models and enhancing the IoT prototype are (AIPA 2012), India, 292-296.
[20] Channe, H., Kothari, S., & Kadam, D. (2015). Multidisciplinary model
future course of this investigation. for smart agriculture using internet-of-things (IoT), sensors, cloud-
computing, mobile-computing & big-data analysis. Int. J. Computer
Technology & Applications, 6(3), 374-382.
R EFERENCES [21] Difallah, W., Benahmed, K., Draoui, B., & Bounaama, F. (2017). Lin-
ear optimization model for efficient use of irrigation water. International
[1] Govt. of India, “Indian economic survey”, 2018. http://mofapp. Journal of Agronomy, 2017.
nic.in:8080/economicsurvey/
[2] Sharma, D., Bhondekar, A. P., Ojha, A., Shukla, A. K., & Ghanshyam,
C. (2016). A Technical Assessment of IOT for Indian Agriculture
Sector. In 47th Mid-Term Symposium on Modern Information and
Communication Technologies for Digital India, Chandigarh.
[3] Chaudhary, R., Pandey, J. R., Pandey, P., & Chaudhary, P. (2015).
Case study of Internet of Things in area of Agriculture,‘AGCO’s Fuse
Technology’s’‘Connected Farm Services’. In Green Computing and
Internet of Things (ICGCIoT), 2015 International Conference on (pp.
148-153). IEEE.
[4] Verma, N. K., & Usman, A. (2016). Internet of Things (IoT): A relief
for Indian farmers. In Global Humanitarian Technology Conference
(GHTC), 2016 (pp. 831-835). IEEE.
[5] Bekiroglu, K., Srinivasan, S., Png, E., Su, R., Poolla, K., and Lagoa, C.
(2017, December). An Internet of Things compliant model identification
methodology for smart buildings. In Decision and Control (CDC), 2017
IEEE 56th Annual Conference on (pp. 4440-4445). IEEE.
[6] Ryu, M., Yun, J., Miao, T., Ahn, I. Y., Choi, S. C., & Kim, J. (2015).
Design and implementation of a connected farm for smart farming
system. In Sensors, 2015 IEEE (pp. 1-4). IEEE.
[7] Kalathas, J., Bandekas, D. V., Kosmidis, A., & Kanakaris, V. (2016).
Seedbed based on IoT: A Case Study. Journal of Engineering Science
& Technology Review, 9(2).
[8] Wenting, H., Zhiping, X., Yang, Z., Pei, C., Xiangwei, C., & Ooi, S.
K. (2014). Real-time remote monitoring system for crop water require-
ment information. International Journal of Agricultural and Biological
Engineering, 7(6), 37-46.
[9] Baranwal, T., & Pateriya, P. K. (2016). Development of IoT based smart
security and monitoring devices for agriculture. In Cloud System and
Big Data Engineering (Confluence), 2016 6th International Conference
(pp. 597-602). IEEE.
[10] Lee, M., Hwang, J., & Yoe, H. (2013). Agricultural production system
based on IoT. In Computational Science and Engineering (CSE), 2013
IEEE 16th International Conference on (pp. 833-837). IEEE.
[11] Deksnys, V., et al. ”Remote agriculture automation using wireless
link and iot gateway infrastructure.” Database and Expert Systems
Applications (DEXA), 2015 26th International Workshop on. IEEE,
2015.
[12] Dan, L. I. U., Xin, C., Chongwei, H., & Liangliang, J. (2015).
Intelligent agriculture greenhouse environment monitoring system based
on IOT technology. In Intelligent Transportation, Big Data and Smart
City (ICITBS), 2015 International Conference on (pp. 487-490). IEEE.
[13] Harun, A. N., Kassim, M. R. M., Mat, I., & Ramli, S. S. (2015, May).
Precision irrigation using wireless sensor network. In Smart Sensors and
Application (ICSSA), 2015 International Conference on (pp. 71-75).
IEEE.
[14] Hari Ram, V. V., Vishal, H., Dhanalakshmi, S., & Vidya, P. M.
(2015, July). Regulation of water in agriculture field using Internet Of
Things. In Technological Innovation in ICT for Agriculture and Rural
Development (TIAR), 2015 IEEE (pp. 112-115). IEEE.

215

View publication stats