Professional Documents
Culture Documents
Notre Dame University - Louaize: Faculty of Engineering Department of Electrical, Computer, and Communication Engineering
Notre Dame University - Louaize: Faculty of Engineering Department of Electrical, Computer, and Communication Engineering
Notre Dame University - Louaize: Faculty of Engineering Department of Electrical, Computer, and Communication Engineering
Faculty of Engineering
Senior Project
AgroVue
Prepared by:
Karina Jaalouk
Thomas Souaid
Advised by:
Fall 2022
AgroVue
By:
Karina Jaalouk
Thomas Souaid
Fall 2022
Advised by:
Committee Member
Dr. Nassar Mendalek
Dr. Elias Nassar
Department of Electrical, Computer,
Department of Electrical, Computer,
and Communication Engineering
and Communication Engineering
ii
Disclaimer
This report is prepared by students of the Faculty of Engineering at Notre Dame University – Louaize, who are
the sole responsible for the integrity of its content. The advisor and committee member act only in an advisory
capacity and cannot be held responsible, under any circumstances, for any eventual misuse or violation of
intellectual property laws.
ii
Abstract
limitations to be gone to draw in a continuous real-time labor force, connect groups, and make
collaboration. It drives interest in technologies to better monitor, control, and protect against
failures, guarantee product satisfaction, increase productivity, and security, and upskill staff
To meet these priorities, we must keep advancing industries' digital transformation and
deploy these key technologies, using a shared digital awareness that enables these key
automated operations are becoming autonomous and the new normal. All of these can either
enter the mainstream or -if it exists- keep gaining acceptance. All connect to the overall
digital transformation of the agriculture industry. Our goal is to improve the agriculture
industry through technology to develop upon the needs generated for farmers to succeed on a
Acknowledgments
We wish to thank Dr. Nassar Mendalek and Dr. Elias Nassar for their valuable insight
and guidance regarding the technical content and formatting requirements of our senior project
report.
We would also like to thank Dr. Lara Khabaz for her assistance in the preparation of the
A special word of thanks goes to the Faculty of Engineering at Notre Dame University
for the effort and dedication it has invested in us in order to provide us with the knowledge and
Table Of Contents
Abstract ...................................................................................................................................... ii
Table Of Contents ..................................................................................................................... iv
List Of Figures .......................................................................................................................... vi
List Of Tables .......................................................................................................................... vii
Chapter 1 – Introduction ............................................................................................................ 1
1.1. Problem Statement ...................................................................................................... 1
References ................................................................................................................................ 33
vi
List Of Figures
List Of Tables
Chapter 1 – Introduction
through extensive internet searching alongside retrieval of Journal and Conference papers
from IEEE. Furthermore, we inducted a site survey that provided valuable insights into the
main challenges that commercial greenhouse agriculture faces. For example, fertilization
alongside pest control and disease management are applicable to maintain the health of the
crops. Weather monitoring conditions are vital to adjust to and protect the crops, where
remarkably high and low temperatures are translated into a significant presence of diseases
and death to plants from frost bites of the wintry weather and the reduction of the absorption
level of water to nutrients from the ground due to the hot weather. Also, the soil type is
identified and maintained to improve the crop outcome. Therefore, crops are affected by
In this chapter, the problem, needs and objective statements are presented that lead to
the understanding of our topic of choice. Moreover, the skills required and acquired during
unpredictable weather conditions, thus farmers collect data manually, which is inefficient,
The planet generates enough food to feed its entire population. However, not everyone
Deloitte, over 30% of the food produced is discarded between harvest and consumption in
our present supply chain for food production [4]. Moreover, by 2025, the world’s
urbanization [16]. Not only is the land being compromised but also the negative effects of
climate change are directly impacting farmers. Crop output is being influenced by elevated
temperatures interrupting growth and causing quality and waste to deteriorate [16]. A
system is needed to control green houses to boost production per area of land and
To use data acquired through digital transformation, farmers must focus on their
methods to provide and utilize information. The data visualization hardware and software
used applications are suitable for many tasks. However, within a connected digital
Farmers and their supply chain demands face risk and operational interruption as they
struggle to minimize their dependence upon suppliers. Due to the inability to fulfill the
high request of consumers, the pandemic further revealed supply chain flaws and
vulnerabilities. Farmers realize that traditional ways are inadequate to compete today, and
3
there is a real need to shift to more adapted, agile, and intelligent tools that are digitally
enabled.
The objective of this project is to design and prototype a system that can control
producing crops in greenhouses can be strictly controlled to minimize waste and boost
• Soil Moisture
• Soil Temperature
All the above is monitored through a remote device to become labor-saving and time
efficient. A valve will be placed on the water tank so that the farmer can control remotely
the water being delivered to the crops as the user sees fit.
When linked with IoT and analytics, visualization tools will offer the potential to
provide users with immersive practices, real-time contextualized data, and role-associated
data to help users execute their regular tasks instantly, precisely, and successfully. It can
improve the user experience and alter the customer interaction standard to enhance the way
implementation, integration and maintenance. Collection of data from various levels of low
algorithms developments and design through machine learning, Internet of Things and
cloud computing.
Other soft skills will be acquired from team player, leadership, time management,
The gain from the overall experience is teaching us the importance of product
management from creation to design and implementation, and testing aligned with clarity
The LoRa communication network protocol follows the regional standards parameters
listed by The lora alliance (an open association of collaborating members) non-profit
organization that regulates it. The frequency range parameters for Lebanon is from 862-
Major backfire on local component market supply. Its cause is due to an unprecedented
economic crisis. The project scope is revised recurringly to meet minimum requirement
compliance.
elements like direct sunshine, wind, temperature, etc., which may cause reading
its distant location. As a result, the gateway won't have access to WiFi or a power source,
Having one of the wireless nodes positioned right adjacent to the irrigation system might
pose a safety and electrical threat in the event of water leaks and the exposure of electronic
equipment to water.
availability. The camera selected in our scope of work is only compatible with Raspberry
Pi microcontroller.
Due to the nature of all electrical components and the unpredictability of their life
duration, some materials and batteries utilized may be subject to sustainability constraints.
Engineering and marketing requirements assigned for the project is shown in the table
below.
Marketing Engineering
Justification
Requirements Requirements
1, 2, 4, 6 The tolerance of every Based upon hardware benchmarking
sensor should not exceed a specifications, sensors have up to ± 2%
5% margin. error margin. Also, external factors such
as the weather (direct sunlight, wind
etc...) will create reading fluctuations.
Marketing Requirements
1. Data Analytics To Prevent Excess Fertilization And Irrigation
2. Operate & Maintain Ideal Microclimate Control
3. Avoid Diseases
4. Minimize Manual Intervention
5. Visualizing Representation
6. Minimize Costs And Increase Return On Investment
7. Provide Real-Time External Weather Interface
8. On Site Monitoring
9. Maintenance And System Availability
1
Machine Learning
8
system that is highly nonlinear and strongly coupled, influenced by the outside weather
[11]. The basic set of variables that characterize the internal climate is the air temperature,
air humidity, and soil measurements [11]. Identifying them as process outputs, consider
Inputs:
• T; internal temperature
• H; internal humidity
Outputs:
Module AgroVue
Inputs • Sensors: temperature, humidity, soil moisture, soil temperature
• Water station: actuator valve control
• Motion sensor
The project architecture is divided into three vertical tiers, as shown in the figure 2
above. The sensors and actuators in the first left layer interact wirelessly with each other
through LoRa wireless communication modules. It breaks down into two nodes: the first
is soil sensors and the second is a weather and water management system.
The second tier is the gateway and the cloud network layer, which is in charge of
The third tier is the application and user layer, which allows the end user to visualize
real-time data, control the actuator, and check the weather forecast.
In addition to traditional methods of collecting data through sensors. Our system design
enables the farmer to download an application on his mobile device, such as Plantix, which
11
is based on deep learning and identifies probable faults and nutrient deficits in the soil. The
recognized image may be integrated into the dashboard we're using through API. Other
techniques include capturing a regular photo and showing it on your dashboard via a cloud-
All nodes feature a LoRa wireless module transceiver, and an antenna interfaced with
all two microcontrollers, allowing the ESP microcontroller and the gateway to
communicate wirelessly. In addition, all the nodes and gateway have a power operation of
The soil monitoring node consists of four moisture sensors, four soil temperature
The four moisture sensors are linked to the microcontroller, and their average is shown
on the dashboard. The soil temperature sensors are connected to one pin of the
microcontroller. Both types of sensors are placed in the soil near the plant's roots to get the
desired reading.
The water management and indoor monitoring node consist of a temperature and
humidity sensor, LED screen, motion sensor, solenoid valve, an Arduino pro mini, a LoRa
wireless module transceiver, and an antenna. The temperature and humidity sensor are
connected to the microcontroller and their reading values are shown on the dashboard. The
motion sensor and the LED screen are connected to the microcontroller, so when the
motion sensor detects movement in the greenhouse the LED screen will turn on, to reduce
power consumption, and the user can check or detect any fluctuations between on-site
nodes and dashboard. The solenoid valve actuator is connected to the same microcontroller.
The LoRa transceivers will communicate with each other and send the sensor values to
the ESP microcontroller. This microcontroller will communicate with a gateway that is
located in a distant location from the greenhouse. It will ensure that the microcontrollers
with the LoRa wireless module transceiver and antennas will manage to connect to a
The dashboard connect to the network service provider and manage two-way
communication with sensors and the actuator. It will display the values of the sensors
In this circuit, an Arduino Pro Mini is being used to monitor the soil moisture and
temperature. It also includes the LoRa transmitter and its antenna. The TX, RX, VCC,
ground pins, and antenna connection are all connected similarly to the camera node.
Moreover, the enable or auxiliary pin is connected to the ground, between two resistors for
safety reasons.
Four soil moisture sensors are used, and the ground pin of each one is connected to the
ground of the microcontroller, consequently, the VCC of each one is connected to the
power supply. This sensor consists of three pins; the third pin of each sensor is connected
14
to an analog input of the microcontroller. The analog inputs used, in our case, are A0, A1,
Four soil temperature sensors are used each with three pins, same as the soil moisture
sensors. The ground pin is connected to the microcontroller's ground, As a result, the VCC
of each is linked to the power supply. While the third pin of each is linked to a 4.7-ohm
The first part, the interior monitoring node, consists of a humidity and temperature
sensor as well as an LED screen. All three components of this section will have their ground
pin linked to the microcontroller's ground, and therefore their VCC pins will be connected
to the power supply. The temperature and humidity sensor has three pins, one of which
will be attached to a digital pin on the microcontroller, D1. The LED screen has four pins,
with pins 1 and 2 each linked to a separate analog pin and digital pin, A0 and D0,
respectively. The IR senor’s 3rd pin is connected to the S3 pin of the ESP8266.
15
The second part is the water management circuit. It consists of a solenoid valve and its
relay. The solenoid valve needs around 12 volts to operate, hence, the separate 12V power
and the relay. For the solenoid valve to properly work, its ground pin will be connected to
the 12V power supply’s ground and its 2nd pin, VCC, will be connected to the 2nd pin of
the relay. Additionally, the 5th and 6th pins of the relay are the ground and VCC pins; the
VCC pin will be connected to the power supply of the circuit, while the ground pin will be
connected to the ground. Finally, the input pin of the relay, the 4th pin, will be connected
to a digital pin, D2, on the ESP8266. The power supply will be connected to the 1st spin of
the relay, N0. This will ensure that the valve will get the proper voltage needed to operate
smoothly and it does not disrupt the 3.3-5 V operating power of the microcontroller.
16
The two end nodes will communicate through the LoRa topology. LoRa operates in the
ISM band and is a spread-spectrum modulation technique. LoRa is derived from Chirp
Spread Spectrum (CSS) technology. Chirp stands for Compressed High intensity Radar
Pulse. A chirp is a ramp from minimal to maximal (up-chirp) or from maximal to minimal
frequency (down-chirp) [10]. LoRa’s CSS technology is energy-efficient, robust and serves
long ranges.
LoRa is working in half duplex mode, which means sending and listening
simultaneously is not possible [17]. Thus, LoRa uses separate messages for sending and
receiving data, namely uplink and downlink messages. Therefore, the node with the
Arduino Pro mini will be the node sending data and the node with the ESP8266 will receive
the data. Via a radio frequency transmitter radio waves are emitted, in our case 868 MHz.
When all the data is transmitted and located and ESP node, the ESP8266 NodeMCU
will be able to connect to Wi-Fi via 2.4 GHz. This will secure a safe and reliable connection
from our end nodes to the Internet. The NodeMCU offers a variety of development
platform based on ESP8266 which can connect objects and let data transfer using the Wi-
microcontrollers [3].
The ESP8266 contains a 2.4Ghz antenna, ESP-12E chip and 802.11b/g/n HT40 Wi-Fi
transceiver, so it can not only connect to a WiFi network and interact with the Internet, but
it can also set up a network of its own, allowing other devices to connect directly to it [5].
3.4.4. Dashboard
A MQTT protocol is being used to transfer the data from the end devices to the end
user. A lightweight open messaging protocol called MQTT (MQ Telemetry Transport)
gives network clients with limited resources an easy way to share telemetry data in low-
The most popular messaging protocol for the Internet of Things is MQTT (IoT). The
protocol is a set of guidelines that specify how Internet of Things (IoT) devices can publish
and subscribe to data [6]. MQTT is used to send and receive messages and data between
Internet of Things (IoT) and Industrial Internet of Things (IIoT) devices, such as embedded
hardware, sensors, industrial PLCs, etc. The protocol uses the publish/subscribe (Pub/Sub)
paradigm to link devices and is event driven. Topics are used for communication between
the sender (Publisher) and the recipient (Subscriber), who are separate from one another
18
[6]. The MQTT broker is in charge of maintaining the link between them. All incoming
messages are filtered by the MQTT broker, who then correctly distributes them to the
Subscribers [6].
Using MQTT, the dashboard will be developed. A dashboard is a tool for data
visualization that monitors, evaluates, and presents metrics, KPIs, and important data
points [12]. Users of all skill levels may utilize dashboards to comprehend business
intelligence and use it to make better decisions. By gathering data, showing trends or
be understood right away, users actively participate in the analytics process [12]. Charts,
tables, and gauges are used to display visible data on dashboards. Dashboards can be used
to monitor progress, track, and identify patterns over time, offer insightful data on
consumer behavior and market developments, and aid in decision-making [1]. They offer
a quick and simple approach to track progress and spot possibilities by gathering important
The dashboard process will be divided into four parts. The first being the monitoring
The above flowchart is the process that every sensor should go through. To start the
greenhouse screen will load, then it will check for any available data, or in other words
available sensors. In case there is no available data, an error message will be displayed,
and the process would end. While in case there is a sensor available, a request to the
microcontroller will be sent to get the data, then the data can be sent to the database.
Converting the digital data into a widget alternative and then displaying it is crucial for the
user experience. The system will check if the data received is new, or in other words, if it
is different from the previous one, if it is not different the system will check if a different
greenhouse was chosen. Now in case, there is new data, it will be incremented on its graph
and displayed on the screen. Finally, the system will check if a different greenhouse was
chosen if the same greenhouse is being displayed, the system will check if there is data
available. If a different greenhouse was chosen to be displayed, then the greenhouse screen
Figure 7 illustrates the flowchart of the valve control function. First, the screen will load
and will check for the user’s change of the valve state. If the user did not change the state
of the valve, then the system will keep checking until a change is present. When the user
changes the state of the valve, the system will check if the valve is present. In case the
valve is not present then a display error message will be displayed, and the process will
end. However, if the valve is present then the change will be converted to a digital value
and sent to the actuator through the microcontroller. Then, the system will check for a
change in the state of the valve, and the process will repeat itself.
22
Figure 8 illustrates the flowchart of the alert system. The widget will load and will enter
the checking process. First, it will check if the soil moisture is less than 60 and will send
an alert and display it on the screen; this indicates that the soil needs water. Second, it will
check if the temperature is less than 12 or greater than 30 and will send an alert and display
it on the screen; this indicates that the temperature of the greenhouse is abnormal, and that
irrigation is not possible. Then, the system will repeat the checking process.
To develop the dashboard for the project, various dashboard services were tested and
examined. However, none are 100% open source and compatible with the project’s IoT
goals. The availability of open-source and proprietary platforms is limited. Also, they are
likely to predominate. Weeks of researching and testing have led to several dashboards to
OpemRemote, Freeboard, Grafana, etc. [16]. Finally, using ThingsBoard, a dashboard was
On the left, a weather forecast is visualized using an API. Under it, it shows the humidity
and temperature inside the greenhouse as well as the moisture and temperature of the soil.
Also, the alerts will be displayed, on the far right, in addition to a switch to turn on and off
the valve. Finally, graphs will be drawn for each of the required value.
Note: the values are not shown on the dashboard because the sensors are not connected.
24
4. 1. Gantt Chart
One of the most common and effective methods of displaying activities displayed
against time is a Gantt chart, which is frequently used in project management [9]. As list
of the activities is located on the chart's left side, and a suitable time scale is located along
the top. A bar is used to symbolize each activity, and the position and length of the bar
correspond to the activity's beginning, middle, and finish dates [9]. Moreover, the
percentage of the completion of the task is also displayed. Figures 10 and 11 represent the
Comparing the anticipated or predicted costs and benefits connected with a project
analysis [18]. In general, cost-benefit analysis entails totaling up the expenses associated
with a project. In Table 3, a cost analysis table was made for this project. Note that all
In this part, the code for the ESP8266 will be explained. In brief it will highlight the
5.1.1. Libraries
The libraries needed for this part is the DHT library for the humidity and temperature
sensor. The ESP8266 Wi-Fi library, which will allow it to connect to Wi-Fi. The
ThingsBoard library which will help us create MQTT clients and publish the sensors’ result
to the dashboard. Also, the LoRa and SPI libraries that will allow the communication
5.1.2. Setup
Setting up the code consists of defining Wi-Fi access point and its password, creating
an object for the Wi-Fi, defining the Token Access Key for the ThingsBoard network,
creating an object for the ThingsBoard server, and defining the inputs of the sensors and
LED screen.
Also, initiating the Wi-Fi connection and setting up the ThingsBoard server will be done
in this step. Beginning the LoRa protocol by setting the frequency at 868 MHz will
5.1.3. Loop
The loop will constantly check if the Wi-Fi is connected, as well as the ThingsBoard
network. Every second the loop will call the functions to get the LoRa data and the
Initializing the ESP8266 will be the first step. In this part the function will check if the
ESP is present and if it has a stable connection to the Wi-Fi access point. It will also
initialize the connection between the ThingsBoard and the ESP using the Token Access
Using the DHT library, it will be possible to directly read the values of the humidity and
temperature. Using the ThingsBoard library, it will be possible to directly send the values
Before receiving the data, the function will check if there is available data. If there is
data, it will read the first packet and save it. Five seconds later, it will read the second
packet and save it. Similarly, to the previous part, it will be possible to send these values
This function will read the IR sensor, if the sensor detects motion, it will turn on the
LED screen. If the LED screen receives positive feedback from the IR sensor, it will turn
on and display the values of the DHT sensor which was previously obtained above. The
In this part, the code for the Arduino Pro Mini will be explained. In brief it will highlight
5.2.1. Libraries
The libraries needed for this part is the One Wire and Dallas Temperature library for the
soil temperature and soil moisture sensor. Also, the LoRa and SPI libraries that will allow
5.2.2. Setup
Setting up the code consists of defining the inputs of the sensors, as well as, calibrating
the sensors by setting constant values when the sensors are in air and when they are in the
soil.
Beginning the LoRa protocol by setting the frequency at 868 MHz will initialize the
LoRa transceiver.
5.2.3. Loop
Every second the loop will call the functions to get the LoRa data and the soil data. This
After reading the values of the sensors, the values will be mapped from zero to hundred
for the humidity percentage and will be saved. Also, the average of all 4 soil moisture
sensor will be saved as the final value. However, the temperature will directly be read and
saved.
Using LoRa Library, a packet will be created and will print the value of the soil moisture
sensor. Then , the packet will end. After 5 seconds, a packet will be created and will print
the value of the soil temperature sensor. Then, the packet will end. After 5 seconds, the
6.1. Conclusion
A final graduation project on IoT in agriculture with a focus on sensors and LoRaWAN
technology could involve the design and implementation of a system that uses sensors and
production. The project could involve the development of sensors for measuring soil
moisture, temperature, and nutrient levels, as well as sensors for tracking the location and
condition of crops and livestock. The data collected by these sensors could be transmitted
to a central server or cloud platform using LoRaWAN technology, which offers long-range,
low-power communication capabilities. This data could then be used to monitor and control
various agricultural processes, such as irrigation and fertilization, as well as to predict and
respond to potential problems, such as pests or diseases. The project could also involve the
easily access and manage the data and control the system. Overall, the aim of the project
could be to demonstrate the feasibility and potential benefits of using sensors and
The idea for the current study was a medium to large-scale greenhouse use case for data
gathering from soil temperature and moisture monitoring to develop methods that can
improve farming and reduce capital expenditure, improve the quality of crops, as a result,
32
and benefit the food market in cities. A prototype for monitoring soil moisture and
Although LoRaWAN has gained popularity lately, LoRaWAN devices are not yet plug-
and-play. Problems were encountered in setting up the use case, such as difficulties
examination of the setup procedure and its technological hazards, a method of simplifying
the setup for IoT application developers utilizing LoRaWAN-based communication can be
provided. This signifies that it is a framework. It helps the sensor application developer
during installation and testing the architecture shown here incorporates a specific prototype
sensing device and links to The ThingsBoard dashboard. This prototype is scalable and can
Furthermore, additional and diverse sensors may be incorporated. A bigger payload will
result in a higher transmission burden and a longer duration on air, which the framework
can monitor. These measurements might help developers make decisions about payload
structure. Another ideal feature would be the ability to serve more than one sensor, at the
same time, which is readily enabled with a few framework modifications. Help with the
setup and testing of a whole sensor network. Another enhancement would be to set up a
References
[2] C. Bernstein, K. Brush, and A. S. Gillis, “What is MQTT and how does it work?,”
IoT Agenda, 27-Jan-2021. [Online]. Available:
https://www.techtarget.com/iotagenda/definition/MQTT-MQ-Telemetry-Transport.
[Accessed: 30-Sep-2022].
[4] K. Szegedi, “Food waste has gone viral,” Deloitte Switzerland, 19-Feb-2021.
[Online]. Available: https://www2.deloitte.com/ch/en/pages/consumer-
business/articles/food-waste-has-gone-viral.html. [Accessed: 20-Sep-2022].
[5] Last Minute Engineers, “Insight into ESP8266 NODEMCU features & using it with
Arduino IDE (easy steps),” Last Minute Engineers, 18-Dec-2020. [Online].
Available: https://lastminuteengineers.com/esp8266-nodemcu-arduino-tutorial/.
[Accessed: 13-Dec-2022].
[6] “Mqtt Essentials,” MQTT Essentials - All Core Concepts explained, 2022. [Online].
Available: https://www.hivemq.com/mqtt-essentials/. [Accessed: 30-Sep-2022].
[7] “Nodemcu ESP8266 specifications, overview and setting up,” Make, 25-Mar-2022.
[Online]. Available: https://www.make-it.ca/nodemcu-details-specifications/.
[Accessed: 14-Dec-2022].
https://www.mobilefish.com/developer/lorawan/lorawan_quickguide_tutorial.html.
[Accessed: 14-Dec-2022].
[12] R. Serpa, “How to build data visualization dashboards,” Sigma Computing, 21-Jul-
2022. [Online]. Available: https://www.sigmacomputing.com/blog/how-to-build-
data-visualization-dashboards/. [Accessed: 20-Sep-2022].
[13] Rp2-1.0.3 LoRaWAN® regional parameters. LoRa Alliance®. (2021, July 20).
Retrieved November 4, 2022, from https://lora-alliance.org/resource_hub/rp2-1-0-3-
lorawan-regional-parameters/
[14] SemTech Corporation , “What are Lora® and Lorawan®?,” LoRa Developer
Portal, Feb-2020. [Online]. Available: https://lora-
developers.semtech.com/documentation/tech-papers-and-guides/lora-and-lorawan/.
[Accessed: 14-Dec-2022].
[15] S. Cope, “Steve's Internet Guide,” |Guide to IOT Dashboards and Platforms, 04-Jul-
2022. [Online]. Available: http://www.steves-internet-guide.com/iot-mqtt-
dashboards/. [Accessed: 27-Oct-2022].
[17] “TR 103 526 - v1.1.1 - System Reference document (SRdoc); Technical
characteristics for Low Power Wide Area Networks Chirp Spread Spectrum
(LPWAN-CSS) operating in the UHF spectrum below 1 GHz,” ETSI, Mar-2018.
[Online]. Available:
https://www.etsi.org/deliver/etsi_tr/103500_103599/103559/01.01.01_60/tr_10355
9v010101p.pdf. [Accessed: 14-Dec-2022].
[18] T. Stobierski, “Cost-benefit analysis: What it is & how to do it: HBS Online,”
Business Insights Blog, 05-Sep-2019. [Online]. Available:
https://online.hbs.edu/blog/post/cost-benefit-
analysis#:~:text=A%20cost%2Dbenefit%20analysis%20is,sense%20from%20a%2
0business%20perspective. [Accessed: 19-Oct-2022].