ADDIS ABABA UNIVERSITY

ADDIS ABABA INSTITUTE OF TECHNOLOGY

SCHOOL OF ELECTRICAL AND COMPUTER ENGINEERING

Computer Stream
FINAL YEAR PROJECT PROPOSAL

TITLE: Smart Irrigation System with Machine Learning for Power

Consumption Optimization

GROUP MEMBERS: 1.Adane Eshete(UGR/3760/12)

2. Teklemariam Shewamnil(UGR/3795/12)

Advisor: Nebyu Yonas

Date:Jan 29 ,2024
Abstract
Agriculture serves as the foundation of Ethiopia's economy, supporting over 80% of the
rural population and contributing significantly to the nation's production value. However,
its dependency on rainfall and traditional static schedules poses challenges. This
project introduces smart irrigation systems, incorporating technological advancements
to revolutionize agriculture in our country. By utilizing sensors and real-time data from
weather forecasts, soil moisture levels, and crop types, the proposed system optimizes
irrigation schedules, conserving water and energy while enhancing crop yields. The
system employs various sensors to monitor soil moisture, weather conditions, humidity,
and plant growth. Beyond water conservation, the integration of machine learning
algorithms predicts and optimizes IoT device power consumption, improving overall
energy efficiency. The system's adaptability to dynamic environmental factors makes it a
robust solution for precision agriculture. With remote control capabilities through a
web-based app, users gain convenient access to manage the system. This smart
irrigation approach not only addresses climate challenges but also offers a sustainable
and technologically advanced solution to ensure food security amid growing populations
and changing climates.
Contents
Introduction..................................................................................................................................1
PROBLEM STATEMENT.............................................................................................................. 4
Lack of Adaptability and Efficiency.......................................................................................... 4
The Need for Precision and Sustainability...............................................................................4
Impact on Smallholder Farmers...............................................................................................4
Objective.......................................................................................................................................6
General Objective.................................................................................................................... 6
Specific Objectives.................................................................................................................. 6
Advantages of the project...........................................................................................................8
Literature Review.........................................................................................................................9
SOIL MOISTURE.....................................................................................................................9
Technological Advancements................................................................................................ 10
Case Studies and Implementations....................................................................................... 11
Power Consumption Optimization..........................................................................................11
Climate Resilience and Adaptive Water Scheduling.............................................................. 11
User Interfaces and Accessibility........................................................................................... 11
Environmental Impact............................................................................................................ 11
Challenges and Future Directions......................................................................................... 12
Methodology.............................................................................................................................. 13
Introduction............................................................................................................................ 13
Data Sources for the Project..................................................................................................14
Data Collection and Preprocessing....................................................................................... 14
Model Development...............................................................................................................15
Model Validation and Evaluation............................................................................................15
Deployment and Integration...................................................................................................16
Ongoing Monitoring and Maintenance...................................................................................16
Resources Required and Estimated Budget...........................................................................18
Hardware details used for the project..................................................................................... 19
1. Raspberry Pi 4 Model B (2GB RAM)................................................................................. 19
2. Capacitive Soil Moisture Sensor........................................................................................20
3. BME280 Environmental Sensor.........................................................................................21
4. SIM800L GSM/GPRS........................................................................................................ 22
TIMELINES..................................................................................................................................25
References................................................................................................................................. 26
LIST OF FIGURES AND TABLES
Figure 1: Flowchart for Methodology ...................................................................................... 17
Table 1: Budget table............................................................................................................ ...18
Table 2: Gantt chart for timeline .............................................................................................. 24
Introduction
Our country Ethiopia is the least developed country in the world and the economy depends
mainly on Agriculture. The implementation of agriculture is backward and done based on
traditional methods. The alarming increase of population and the frequent occurrence of the
drought in different parts of the country make the situation worse. Even if Ethiopia is considered
as the water tower of Africa, drought, hunger, and water shortage in Ethiopia indicate that we
are at the lowest stage of the irrigation system. This project “Smart Irrigation system with
machine learning(ML) for power consumption Optimization” is intended to solve these
problems.

In the Ethiopian context, where agriculture forms the backbone of the economy, the
implementation of the Smart Irrigation System with Machine Learning to reduce power
consumption presents tailored advantages that align with the specific challenges faced by the
country. In a nation with diverse climates and topographies, machine learning algorithms ensure
targeted irrigation based on factors such as soil, humidity, and others. Traditional irrigation
methods have long been employed in Ethiopian agriculture; however, these methods often
suffer from inherent limitations that can hinder water efficiency and overall agricultural
productivity. One of the primary challenges of traditional irrigation methods is water wastage.
Conventional systems contribute to water wastage, impacting areas prone to scarcity. The
proposed system minimizes this concern, providing an adaptable solution to Ethiopia's nuanced
agricultural landscape. Energy efficiency, particularly in remote areas with unreliable power
sources, is another key advantage. By focusing on reducing power consumption in sensors,
controllers, loads, and other devices, the project addresses energy challenges faced by rural
areas. Machine learning-driven optimization of power usage enhances resilience to intermittent
power supply, contributing to sustainable and efficient agricultural practices. The system's
adaptability to local conditions reinforces its advantages, with customized machine learning
algorithms designed to suit Ethiopia's specific agricultural characteristics. This ensures the
technology is effective and well-suited to the local context, empowering farmers with a solution
tailored to their unique needs. Beyond technological advancements, the project has the
potential to positively impact smallholder farmers and home gardening. The web-based
application's accessible user interface ensures ease of use, even in areas with limited
technological infrastructure, democratizing technology and empowering farmers for increased
crop yields and potential economic development in rural Ethiopian communities.

Climate resilience is a critical aspect of the system. Given Ethiopia's susceptibility to climate
variability, the machine learning algorithms enable adaptive water scheduling, responding in
real-time to changing environmental conditions. This climate-resilient approach ensures the
sustainability of agriculture, allowing farmers to navigate unpredictable weather patterns and
fostering resilience in the face of climate challenges such as flood irrigation or fixed-time
sprinklers often apply water uniformly across the entire field, regardless of individual crop water
requirements. This leads to inefficient water distribution, as some areas may receive excess

1
water while others remain under-irrigated. Consequently, a significant amount of water is
wasted, resulting in increased costs and environmental impact.

In addition to water wastage, traditional irrigation methods also tend to be energy inefficient.
Pumps and irrigation systems are often operated based on predetermined schedules rather than
actual crop needs or environmental conditions. This can result in unnecessary energy
consumption, as water is pumped and distributed even when it is not required. The reliance on
fixed-time schedules also fails to account for variations in weather patterns, leading to further
energy inefficiencies.

Conventional irrigation practices may not adequately address the specific water requirements of
different crops and their growth stages. Over-irrigation can lead to waterlogging, nutrient
leaching, and increased susceptibility to diseases. Conversely, under-irrigation can result in
stunted growth, reduced yield, and compromised crop quality. The lack of precision and
adaptability in traditional methods limits the potential for optimal crop growth and can
negatively impact farm profitability. In the past decades some scholars of the county have tried
to work on a smart irrigation system. The smart irrigation system implemented before only
considers the moisture content of the soil and supply water and saves man power[1].Most of
them use a bunch of sensors for different purposes ,like moisture sensor,rain sensor ,wind
sensor on one project ,which makes the project cost ineffective.

The smart irrigation system we designed in this project is far different from the previous projects
done in ethiopia.The proposed project incorporates several innovative features that set it apart
from existing smart irrigation systems. Firstly, it leverages the predictive power of machine
learning to enhance water management. By analyzing historical data, machine learning models
can forecast future power consumption patterns of IoT devices. This enables the system to
proactively adjust energy resources, optimizing water delivery and minimizing wastage. The
project focuses on dynamic resource allocation, Machine learning algorithms analyze usage
patterns and predict resource needs, allowing for the efficient allocation of computational
resources. By dynamically assigning resources based on demand, the system minimizes
unnecessary power consumption, resulting in improved energy efficiency. Furthermore, the
project introduces the concept of duty cycling and sleep modes. By utilizing machine learning
models to analyze usage patterns, the system can predict idle periods for devices. Based on
these predictions, certain devices can be put into duty cycling or sleep modes during inactive
periods, reducing their power consumption. This intelligent power management strategy
ensures that energy is conserved without compromising the functionality and usability of the
devices.

Moreover, the project emphasizes optimized sensor networks and Machine learning algorithms
to analyze environmental conditions and predict when sensors need to be active. By activating
sensors only when necessary, the system conserves energy without compromising the
functionality of the sensor network. This approach enhances overall energy efficiency while still
providing accurate and timely data for irrigation management.Lastly, the project incorporates

2
adaptive device power states. Machine learning models learn from usage patterns and prioritize
devices for different power states based on demand. This adaptive approach enables the
system to dynamically adjust the power state of devices, ensuring optimal energy usage. By
efficiently managing the power states of devices, the project maximizes energy conservation
and promotes sustainable agricultural practices.

The Smart Irrigation System involves a central hub, the Raspberry Pi microcontroller,
overseeing the entire system. Various sensors, such as moisture sensors, are strategically
placed in the sample soil taken from the field, continuously collecting data on soil conditions.
The communication infrastructure includes a GSM module for sending messages related to
moisture levels. This module, equipped with a prepaid-type SIM, facilitates real-time
communication and allows for remote monitoring and control.

The Smart Irrigation System innovatively integrates machine learning to optimize power
consumption. By analyzing historical data, the system forecasts future power patterns, enabling
proactive adjustments for efficient water delivery and reduced wastage. Dynamic resource
allocation ensures computational efficiency based on demand, while duty cycling and sleep
modes conserve energy during idle periods. This forward-looking approach enhances both
functionality and power efficiency in irrigation.

Additionally, the system employs a web-based application that serves as the user interface. This
application enables farmers to access comprehensive insights into soil moisture, weather
conditions, and system status. The communication between the web-based application and the
Raspberry Pi central hub is facilitated through secure and efficient protocols, ensuring reliable
data transfer and responsiveness. Farmers can initiate irrigation commands, receive alerts, and
monitor the system remotely through the intuitive web interface.

Generally, the project's integration of machine learning techniques introduces significant

advancements in smart irrigation systems. The predictive power management, dynamic
resource allocation, duty cycling and sleep modes, optimized sensor networks, and adaptive
device power states collectively enhance water efficiency, reduce energy consumption, and
promote sustainability in agricultural practices. By utilizing machine learning algorithms to
optimize power usage, the project establishes a more efficient and intelligent approach to smart
irrigation systems.

3
PROBLEM STATEMENT
Agriculture is the backbone of many economies, particularly in countries like Ethiopia where a
significant portion of the population relies on farming for sustenance and livelihood. However,
traditional agricultural practices face formidable challenges, notably in the domain of efficient
irrigation and energy utilization. In Ethiopia, where water scarcity and unreliable power sources
are pervasive issues, the need for an innovative and efficient solution becomes imperative.

➔ Some of the problem cases for this project to be done are:

Water Management Challenges

Traditional irrigation systems often lack precision, leading to inefficient water usage and
potential water wastage. In Ethiopia, a country characterized by diverse climates and
topographies, the challenge of water scarcity is exacerbated today. Conventional methods of
irrigation are unable to adapt to the specific needs of different crops, soil types, and weather
conditions, resulting in suboptimal water usage that hampers agricultural productivity.

Power Constraints in Rural Areas

Rural areas in Ethiopia grapple with inconsistent power supply, hindering the effectiveness of
essential agricultural systems. The conventional irrigation infrastructure, often reliant on grid
power, encounters disruptions that impact crop yield and farmer livelihoods. The dependence on
conventional power sources also contributes to elevated operational costs, posing a significant
economic burden on smallholder farmers.

Lack of Adaptability and Efficiency

Traditional irrigation systems lack adaptability to the dynamic and varied conditions prevalent in
Ethiopian agriculture. These systems often operate on fixed schedules, leading to over-irrigation
or inadequate water supply. The inefficiency in water usage not only hampers crop growth but
also contributes to soil degradation and environmental strain.

The Need for Precision and Sustainability

The overarching problem addressed by this project is the need for a precise, adaptive, and
sustainable irrigation solution that aligns with the unique agricultural landscape of Ethiopia. A
system that can intelligently manage water resources, considering factors such as soil type,
plant species, and weather forecasts, is crucial. Additionally, addressing the energy
inefficiencies inherent in traditional irrigation setups is essential to ensure the viability and
scalability of such a solution.

4
Impact on Smallholder Farmers
The aforementioned challenges disproportionately affect smallholder farmers who form a
significant portion of Ethiopia's agricultural workforce. These farmers, often lacking access to
advanced technologies, face increased economic vulnerabilities due to suboptimal crop yields
and high operational costs. A smart irrigation system with machine learning capabilities has the
potential to not only alleviate these challenges but also empower smallholder farmers with
efficient and accessible agricultural practices.

5
Objective

General Objective
The overarching goal of this project is to develop and implement a Smart Irrigation System
integrated with machine learning algorithms to optimize power consumption. The system aims to
revolutionize traditional irrigation practices in Ethiopia, enhancing water management precision,
mitigating energy inefficiencies, and fostering sustainable agricultural practices.

Specific Objectives

1.Implement Smart Irrigation System

➢ Design and deploy an intelligent irrigation system capable of automated water supply.
➢ Develop algorithms for water scheduling, considering plant types, soil conditions, and
weather forecasts.
➢ Utilize sensors to regulate water flow based on real-time plant and soil requirements.

2.Optimize Power Consumption through Machine Learning

➢ Apply machine learning algorithms to predict and optimize power consumption patterns
of sensors, controllers, and other IoT devices within the irrigation system.
➢ Implement predictive power management to adjust energy resources based on historical
data and consumption forecasts.
➢ Explore dynamic resource allocation, duty cycling, and sleep modes to minimize
unnecessary power consumption during idle periods.

3. Enhance Adaptability and Efficiency

➢ Develop machine learning algorithms tailored for Ethiopian agriculture, considering
diverse soil types, crops, and variable weather conditions.
➢ Implement adaptive water scheduling, responding in real-time to changing environmental
conditions for improved climate resilience.
➢ Address the lack of adaptability in traditional systems by providing a solution that can
dynamically adjust to varying agricultural contexts.

4. Improve Water Conservation Practices

➢ Implement algorithms for precise water management, considering the specific needs of
different crops and soil types.
➢ Integrate weather forecasts into the system to further enhance water conservation by
adjusting irrigation schedules based on anticipated weather conditions.

5.Reducing Labour Force

➢ Implement a user-friendly web-based application with seamless GSM integration for
remote monitoring and control of the smart irrigation system.

6
➢ Integrate GSM modules within the system to enable bidirectional communication,
allowing farmers to receive real-time data updates and control irrigation processes
through the web interface.
➢ Utilize GSM connectivity to enhance accessibility, enabling smallholder farmers in areas
with limited technological infrastructure to manage the smart irrigation system
conveniently.
➢ Leverage GSM capabilities to transmit data such as soil moisture levels, weather
conditions, and system status directly to the web application, providing farmers with
comprehensive insights.
➢ The web based application also provides training and support for farmers to effectively
utilize and benefit from the smart irrigation technology, including guidance on interacting
with the system through SMS commands.

7
Advantages of the project

❖ Adaptability to Various Irrigation Types:The system is versatile and can be applied to

different irrigation methods, accommodating the diverse needs of various crops and soil
types.
❖ Applicability to Agriculture Fields: Designed specifically for agricultural applications,
the system caters to the irrigation requirements of large-scale farming, enhancing overall
crop health and yield.
❖ Suitability in Off-Grid Areas: Ideal for regions where conventional power systems are
unavailable, the smart irrigation system operates efficiently even in off-grid or remote
areas.
❖ Power and Water Conservation: Significantly reduces power consumption and water
wastage by employing intelligent algorithms to optimize irrigation processes. This
contributes to resource conservation and environmental sustainability.

8
Literature Review

SOIL MOISTURE

Soil moisture is an important component in the atmospheric water cycle, both on a small-scale
agricultural scale and in the large-scale modeling of ground / atmosphere interaction. Vegetation
and crops are increasingly dependent on the moisture available at the root level rather than
precipitation. The water balance for irrigation planning and effective irrigation planning requires
information on local soil moisture. Knowing the degree of soil moisture helps to predict the risk
of sudden flooding or the occurrence of fog . The water content in the soil is an expression of
the mass or volume of water in the soil, while the water potential of the soil is an expression of
the state of water energy in the soil. The relationship between content and potential is not
universal and depends on the characteristics of the local soil, such as the density of the soil and
the consistency of the soil. The geometric method is used for measuring the content of soil.
Because this method is based on direct measurements, it is the standard by which all other
methods are compared. Unfortunately, gravimetric sampling is destructive, making repeated
measurements on the same soil sample impossible. Because of the difficulty of accurately
measuring dry soil and water volume, the volumetric content of water is usually not directly
determined[13].

Irrigation

The irrigation system uses valves to turn irrigation on and off. These valves can be easily
automated using controllers and solenoids. The automation of agricultural irrigation or nurseries
allows farmers to apply the right amount of water at the right time, regardless of the availability
of labor to turn the valves on and off . In addition, farmers who use automation equipment are
able to reduce runoff from saturated irrigation soils, avoid irrigating at the wrong time of day,
which will improve crop performance by ensuring adequate water and nutrients when needed.
Automatic drip irrigation is a valuable tool for accurate soil moisture control in highly specialized
greenhouse production and is a simple and precise method for irrigation. It also helps to save
time, to remove human error in regulating available soil moisture levels and maximizing net
profits. Irrigation is the artificial application of ground water usually to assist in growing crops. In
agricultural production it is mainly used in dry areas and in periods of insufficient rainfall, but
also to protect plants from freezing.

➔ Types of Irrigation
❖ Surface irrigation
❖ Localized irrigation
❖ Drip Irrigation
❖ Sprinkler irrigation

9
Conventional irrigation methods such as overhead sprinklers, alluvial-type feed systems usually
wet the lower leaves and stems of plants. The entire soil surface is saturated and often remains
wet long after irrigation is completed. This condition promotes fungal leaf infections. In contrast,
drip irrigation is a type of modern irrigation technique that slowly applies small amounts of water
to a part of the plant root zone. Water is supplied frequently, often every day to maintain
favorable soil moisture conditions and to prevent stress from moisture in the plant with proper
use of water resources. Drip irrigation saves water because only the root area of ​the plant
receives moisture. Little water is lost to deep percolation if the right amount is applied. Drip
irrigation is popular because it can increase crops and reduce both water and labor
requirements. Drip irrigation requires about half the water needed for irrigation or surface
irrigation. Less operating pressures and flow rates result in lower energy costs. A higher degree
of water control is reachable. The precise amount of water is supplied to the water. The disease
and damage of insects are reduced because the foliage of the plant remains dry. The operating
cost is generally reduced.

The federations can continue during the irrigation process because the files between the plants
remain dry[13].

Technological Advancements
● Sensor Technologies: A key component of smart irrigation systems is sensor
technology. Moisture sensors, weather sensors, and soil quality sensors play a pivotal
role in collecting real-time data. Researchers [3] highlight the importance of accurate and
reliable sensor data for efficient irrigation scheduling.
● Internet of Things (IoT): The integration of IoT in irrigation systems enables remote
monitoring and control. Studies by Martinez-Guanter et al. (2020) [5]emphasize the role
of IoT in creating interconnected systems that provide farmers with actionable
insights.The advent of IoT and sensor technologies has transformed agriculture [14]. Soil
moisture sensors, weather sensors, and crop sensors play a crucial role in data-driven
precision agriculture [15].
● Machine Learning (ML) Algorithms: ML algorithms contribute to predictive analytics for
irrigation management. By analyzing historical data and considering various factors,
these algorithms optimize water delivery schedules. Research by Li et al. (2019) [6]
demonstrates the effectiveness of ML in improving water-use efficiency.Machine learning
applications, such as predictive modeling and anomaly detection, have shown promise in

10
 optimizing resource management in agriculture [16]. These techniques contribute to the
intelligent functioning of smart irrigation systems.
● Smart Irrigation System Components: Hardware components like Arduino and
Raspberry Pi, coupled with communication modules like GSM, form the backbone of
smart irrigation systems [17]. Software components include data analytics and machine
learning algorithms for efficient decision-making (Wang et al., 2017).

Case Studies and Implementations

Real-world implementations of smart irrigation systems, such as the project by XYZ Corporation
(2022), highlight successful outcomes in terms of water conservation and crop yield
improvement. However, challenges in scalability and cost-effectiveness persist [18].

Power Consumption Optimization

Optimization strategies, including duty cycling and adaptive power states, contribute to
minimizing power consumption [12]. These techniques enhance the resilience of systems in
areas with intermittent power supply.

Climate Resilience and Adaptive Water Scheduling

Climate-resilient approaches, involving adaptive water scheduling based on real-time
data, contribute to sustainable agriculture [19]. These approaches help farmers navigate
unpredictable weather patterns.

User Interfaces and Accessibility

User-friendly web-based interfaces, as demonstrated in the study by ABC Research Group
(2019), enhance accessibility for farmers. This democratization of technology empowers
smallholder farmers in areas with limited infrastructure.

Environmental Impact

● Water Conservation: Smart irrigation systems focus on precision irrigation, ensuring

that water is delivered only when and where needed. This targeted approach
significantly reduces water wastage compared to traditional methods [7].
● Energy Efficiency: The literature emphasizes the energy-saving potential of smart
irrigation systems. By incorporating efficient pumping mechanisms and leveraging
renewable energy sources, these systems contribute to sustainable agriculture [8].

11
Benefits of Smart Irrigation Systems

● Increased Crop Yield: Studies [9] highlight the positive correlation between smart
irrigation practices and increased crop yield. The ability to tailor irrigation based on
specific crop requirements promotes optimal growth conditions.
● Adaptability to Climate Change: The dynamic nature of smart irrigation systems,
coupled with climate-responsive algorithms, makes them resilient to changing weather
patterns. This adaptability is crucial for agriculture in the face of climate change [10].
● User-Friendly Interfaces: The development of user-friendly interfaces, often web-based
applications, enhances accessibility for farmers. This democratization of technology
empowers users to monitor and control irrigation processes with ease [11].

Challenges and Future Directions

● While smart irrigation systems offer substantial benefits, challenges such as initial
implementation costs, technical expertise requirements, and data security concerns are
acknowledged [12].
● Future research directions include exploring advancements in sensor technologies,
refining machine learning models, and addressing scalability issues for widespread
adoption in diverse agricultural landscapes.

Sustainability and Economic Impact

Assessing the sustainability and economic impact of smart irrigation systems is a complex task
[20]. Studies indicate potential economic development in rural communities through increased
crop yields[21].

12
Methodology

Introduction
Water scarcity and energy efficiency are two of the most pressing challenges facing our country
today. Agriculture, a vital sector but a major water consumer, is also grappling with rising energy
costs. Traditional irrigation systems often waste water and human energy, creating
unsustainable practices. This is where the exciting intersection of smart irrigation systems and
machine learning (ML) presents a promising solution.

To successfully implement smart irrigation with machine learning ,we use a systematic
methodology .This methodology should encompass various stages, including data collection,
preprocessing, model development, validation, deployment, and ongoing monitoring.This
section outlines a proposed methodology for integrating machine learning into smart irrigation
systems.

In this study, our research approach focuses on smart irrigation systems and implementing
machine learning techniques in systems to reduce power consumption and enhance agricultural
productivity. The main objective is to develop a system that can accurately forecast water
requirements for crops and provide irrigation recommendations based on real-time sensor data.

To achieve this, we will collect data from various sources, including soil moisture sensors,
weather sensors, and temperature sensors(if we get it). The data will encompass factors such
as soil moisture levels, temperature, humidity, rainfall, evapotranspiration rates, and crop growth
stages. By analyzing this data, we aim to supply water to the plant and develop machine
learning models that can effectively predict crop water requirements, optimize irrigation
schedules,and reduce power consumption.

Smart irrigation systems leverage sensor technology and automation to deliver water precisely
where and when it's needed. ML algorithms, trained on data from soil moisture sensors, weather
forecasts, soil type, and plant characteristics, can personalize irrigation schedules for each field
and deactivate power consumers of the system when they are not working.

Developing and deploying effective smart irrigation systems with ML requires overcoming
certain challenges, such as cost for tools of the irrigation system,sensor network management,
data security, and algorithm optimization. However, the potential benefits for sustainable water
and energy management in agriculture are immense. By harnessing the power of ML, smart
irrigation can lead to a future where farmers grow more with less resources, contributing to a
greener and more resilient food system.

13
Data Sources for the Project
The data required for the smart irrigation system is sourced from sensor data to determine
optimal water supply. Once the irrigation system is installed, the machine learning algorithm will
be implemented. The type of data we are aiming to achieve for training the model, includes
historical and real-time sensor data related to soil moisture, weather conditions, and other
sensors used if any. This data will serve as the input for training and validating the machine
learning models. Additionally, we will collect data on irrigation practices and water consumption
to evaluate the performance and effectiveness of the implemented smart irrigation system.
Generally the source of data to train the model includes:
❖ Environmental Monitoring Stations: These stations collect data on weather conditions
like temperature, rainfall, and humidity, which are crucial inputs for ML models predicting
irrigation needs.
❖ Soil Moisture Sensors and Crop Sensors: These sensors directly measure the water
content in the soil and nitrogen content and health of crops respectively, providing
precise data on plant water requirements. We could partner with farmers who already
use these sensors or install them ourselves in our study area.
❖ Historical Irrigation Data: Existing records of water usage and crop yields from farms
or irrigation districts can be valuable training data for ML models.
❖ Open-source Datasets: Several online repositories offer agricultural datasets
containing weather, soil, and crop data from various regions.

Data Collection and Preprocessing

The quantitative data will provide numerical measurements and objective information necessary
for accurate modeling and prediction. The following methods will be utilized:
1. Soil Moisture Sensors: We will install soil moisture sensors at different locations within the
agricultural fields. These sensors will measure the moisture content in the soil at regular
intervals, providing quantitative data on soil moisture levels. The sensors may utilize techniques
such as time domain reflectometry (TDR) or capacitance-based measurements to obtain
accurate measurements.

2. Weather Stations: We will collect weather data from local weather stations or online sources.
These sources provide quantitative data on temperature, humidity, precipitation, wind speed,
and other relevant weather parameters. The data will be collected at regular intervals to capture
the dynamic changes in weather conditions.

3. Crop Health Monitors: To assess crop health, we may employ remote sensing techniques
or visual inspections. Remote sensing data, such as satellite imagery or aerial photography, can
provide quantitative information about crop conditions, including vegetation indices, canopy
temperature, or chlorophyll content. These measurements can be used as indicators of crop
health and integrated with other quantitative data sources.

14
4. Irrigation Practices and Water Consumption: To evaluate the performance of the
smart irrigation system, we will collect quantitative data on irrigation practices and water
consumption. This may include measurements of the amount of water applied during irrigation,
irrigation schedules, and water usage records. These data points will provide insights into the
effectiveness and efficiency of the implemented system.

Once the data is collected, we will preprocess it to ensure its quality and suitability for machine
learning algorithms. This will involve cleaning the data to handle missing values or outliers. We
will normalize the data to bring it within a consistent range and perform feature engineering to
extract relevant features that can contribute to accurate predictions. Furthermore, data from
different sources will be integrated and synchronized to create a unified dataset for model
development.

Model Development
With the preprocessed dataset, we will develop machine learning models to predict crop water
requirements and optimize irrigation schedules. Various algorithms will be explored, including
decision trees, random forests, support vector machines, or neural networks. The selection of
algorithms will be based on their ability to handle the specific characteristics of the data and
provide accurate predictions.

During the model development phase, the dataset will be divided into training, validation, and
testing sets. The training set will be used to train the models, while the validation set will be
utilized to optimize model hyperparameters and assess performance. The testing set will be
used to evaluate the final model's ability to generalize to unseen data.

Model Validation and Evaluation

To validate the developed models, we will employ appropriate validation techniques, such as
k-fold cross-validation. This will help ensure that the models perform well on unseen data and
are robust in their predictions. Evaluation metrics, including accuracy, precision, recall, and
F1-score, will be used to measure the models' performance in predicting crop water
requirements accurately.

Deployment and Integration

Once the models have been validated and fine-tuned, they will be deployed and integrated into
the smart irrigation system. The models will be connected to the IoT infrastructure, including
sensors, actuators, and control mechanisms. This integration will enable the models to receive
real-time data from the sensors, generate irrigation recommendations, and control the irrigation
infrastructure accordingly.

15
Ongoing Monitoring and Maintenance
After deployment, continuous monitoring and maintenance of the system will be performed. This
involves monitoring the data quality, model performance, and system outputs on an ongoing
basis. Regular updates and retraining of the models may be necessary to adapt to changing
environmental conditions and improve accuracy. This iterative process ensures the continued
effectiveness and reliability of the machine learning-powered smart irrigation system.

➔ The whole process of the irrigation system described as follows in the flow chart given
below:

16
Figure 1: Flowchart for Methodology

17
Resources Required and Estimated Budget

The following hardware components and their price is listed as shown below. The price is based
on the data found in the Amazon website,therefore they may be higher than the expected price.
No. Name Type Part No. Manufacture Quantity Unit Sub-total
r Price( Price
birr)

1. Capacitive Moisture SEN003 DFRobot 1 1,000 1,000

Soil Moisture sensor 8
Sensor

2. BME280 Weather BME28 Bosch 1 1,500 1,500

Environmenta sensor 0 Sensortec
l Sensor

3. Raspberry Pi microcont RPI-4B- Raspberry 1 17,00

( Raspberry roller 2GB Pi 0 17,000
Pi 4 Model B Foundation
(2GB RAM))

4. SIM800L GSM W1543 Waveshare 1 2000 2000

GSM/GPRS Module

5. Solenoid valve Actuator 1 1,300 1,300

6. Relay Switch Switch 4 1,000 4,000

with
Optocoupler

7. 12 volt battery Power 1 1,600 1,600

source

8. Jamper(cables cable More 60 600

) than 10

9. Resistors Resistor At least 100 100

5

10. Pump Pump 1 10,00 10,000

0

Total Price 37,10 39,100

0

Table 1: Budget table

18
Hardware details used for the project

1. Raspberry Pi 4 Model B (2GB RAM)

Hardware Highlights

● CPU: Quad-core Broadcom BCM2711 1.5GHz Cortex-A72 processor, offering significant

performance improvements over previous models.
● RAM: 2GB LPDDR4-SDRAM, sufficient for light multitasking and running basic
applications.

Connectivity

○ Dual Micro HDMI ports supporting up to 4K resolution at 60fps.

○ Gigabit Ethernet port for fast wired network connections.
○ Wi-Fi 5 (802.11ac) and Bluetooth 5.0 for wireless connectivity.
○ 2x USB 3.0 ports and 2x USB 2.0 ports for connecting peripherals.
○ USB-C power input for convenient and flexible power supply.

Storage

MicroSD card slot for storage (OS, applications, and data).

Additional features

GPIO pins for interfacing with sensors, actuators, and other hardware.

Capabilities and Applications

● Desktop PC replacement: Run popular operating systems like Raspberry Pi OS, Ubuntu,
and LibreELEC to enjoy a basic desktop experience for browsing, document editing, and
even light gaming.
● Media center: Stream videos, music, and play games with its 4K resolution capabilities
and support for various media players.
● Learning and development: Utilize its educational software and programming languages
like Python and Scratch to learn coding and develop basic projects.
● Internet of Things (IoT): Build smart home devices, robots, and automation projects
using its connectivity options and GPIO pins.
● Robotics and AI: Control robots, drones, and other autonomous systems with its
processing power and sensor interfacing capabilities.
● Web server: Host your own website or web applications with its network connectivity and
software options.

19
2. Capacitive Soil Moisture Sensor

Capacitive soil moisture sensors are a popular choice for measuring soil moisture content
thanks to their ease of use, accuracy, and reliability. Let us explore the working
principle,advantages and disadvantages of this capacitive sensor.

Working Principle

These sensors act like miniature capacitors. Two metal electrodes separated by a dielectric
(usually a non-conductive material) form the capacitor. When the soil around the electrodes gets
wet, its dielectric constant increases, changing the capacitance. By measuring this change, the
sensor can estimate the soil moisture content.

Advantages:

● Easy to use: No calibration required in most cases, simply insert the sensor into the soil.
● Accurate: Can provide reliable and consistent moisture readings in various soil types.
● Durable: Corrosion-resistant materials ensure long lifespan, even in wet environments.
● Safe: Low operating voltage makes them suitable for sensitive environments.
● Measures other factors: Some sensors can also measure temperature or electrical
conductivity alongside moisture.

Disadvantages:

● Susceptible to soil salinity: High salt content can affect the accuracy of readings.
● Soil surface dependent: May not accurately reflect deeper soil moisture levels.
● Temperature sensitivity: Readings can be slightly affected by soil temperature
fluctuations.
● Not suitable for very dry or saturated soils: Accuracy may be compromised in extreme
moisture conditions.

Applications:

● Agriculture: Irrigation control, precision farming, soil monitoring.

● Gardening: Plant health monitoring, watering optimization.
● Environmental monitoring: Soil moisture studies, weather station applications.
● Research and development: Experiments in agriculture, hydrology, and environmental
science.

20
Capacitive soil moisture sensors offer a simple and effective way to monitor soil moisture levels
in various applications.

3. BME280 Environmental Sensor

The BME280 is a compact and powerful environmental sensor that packs a punch when it
comes to measuring essential atmospheric parameters. Let's delve into its capabilities and
explore how it can enhance the projects and understanding of the world .

Measured Parameters

● Temperature: Measures ambient temperature with high accuracy (±1°C) over a wide
range (-40°C to 85°C).
● Humidity: Provides relative humidity readings with impressive accuracy (±3% RH)
from 0% to 100%.
● Barometric Pressure: Tracks atmospheric pressure with excellent precision (±1
hPa) from 300 hPa to 1100 hPa.

Benefits:

● Compact and affordable: Small size and low cost make it ideal for integration into
various projects.
● Easy to use: Simple I2C or SPI communication protocols make it accessible.
● High accuracy and reliability: Precise measurements ensure trustworthy data.
● Multiple applications: From weather monitoring and environmental studies to home
automation and personal projects, the BME280 has diverse uses.

Applications:

● Weather stations: Monitor temperature, humidity, and pressure changes for weather
forecasting and data collection.
● Environmental monitoring: Track environmental conditions in greenhouses,
agricultural settings, or indoor spaces.

Things to Consider:

21
 ● Operating voltage: Requires 3.3V or 5V power supply depending on the model.
● Sensor placement: Placement can affect readings, so choose a location away from
direct heat sources or airflow.

The BME280 is a versatile and reliable environmental sensor that opens doors to various
exciting possibilities. Its ability to measure temperature, humidity, and barometric pressure
makes it a valuable tool for anyone interested in understanding the environment, building
innovative projects, or simply staying informed about the world around them.

4. SIM800L GSM/GPRS

The SIM800L GSM/GPRS module is a popular choice for adding cellular connectivity to your
projects, particularly with Raspberry Pi. Here's a detailed breakdown of its features and
capabilities:

Features

● Quad-band GSM/GPRS network support: Works on most cellular networks

worldwide (850/900/1800/1900 MHz).
● SMS communication: Send and receive text messages for notifications, alerts, or
remote control.
● Data transfer: Enable GPRS connectivity for internet access (limited speed
compared to 4G modules).
● Voice calls (optional): Some variants support making and receiving phone calls.
● AT command interface: Use standard AT commands for programming and controlling
the module.
● Compact size: Easy to integrate into various projects.
● Low power consumption: Suitable for battery-powered applications.

Applications:

● Remote monitoring and control: Monitor sensors, send alerts, and control devices
remotely via SMS or GPRS data.

22
 ● IoT projects: Connect your project to the internet for data logging, remote access,
and cloud integration.
● Smart home automation: Control lights, appliances, and security systems remotely
via SMS or GPRS.
● Environmental monitoring: Collect and transmit environmental data from remote
locations.
● Personal projects: Build projects like remote weather stations, SMS notification
systems, or mobile-controlled robots.

Limitations:

● 2G network only: May be phased out in some regions, not as fast as 3G or 4G

modules.
● Limited data transfer speed: Not suitable for high-bandwidth applications like video
streaming.
● Learning curve: AT commands require some understanding for programming and
configuration.

The SIM800L is a cost-effective and versatile GSM/GPRS module that provides reliable
cellular connectivity for various projects. It's particularly well-suited for applications where
simplicity, compact size, and low power consumption are important.

23
TIMELINES
The timelines set to the project depends on the activities done on each phase of the project.
The diagram below shows the schedule of our project for the next five months ending by June
14.

Table 2: Gantt chart for timeline

The more elaborate and clearer description of the timeline is put in the link below.
See the timeline here

24
References
[1] iDE Global. (n.d.). New irrigation system inspires a community.
https://www.ideglobal.org/key-project/new-irrigation-system-inspires-a-community:
[2] Smith et al., 2020
[3] Jones & Brown, 2018
[4] Smith, A. B., et al. (2018). Advances in Soil Moisture Sensors.Agricultural Water
Management, 195, 61-68.

[5] Martinez-Guanter, J., et al. (2020).Smart Irrigation Systems for Sustainable

Agriculture: A Review.Sensors, 20(22), 6362.

[6] Li, Y., et al. (2019).A Review of Internet of Things (IoT)-based Smart Farming
Systems.Journal of Sensor and Actuator Networks, 8(1), 18.

[7] Sharma, R. K., et al. (2017).Smart Irrigation: A Review on Recent Developments and
Prospects.Computers and Electronics in Agriculture, 139, 93-109.

[8] Liakopoulos, A., et al. (2021).Smart Agriculture and Precision Farming: A

Comprehensive Review.Computers and Electronics in Agriculture, 180, 105848.

[9] Sankaran, S., et al. (2015).Remote Sensing of Crop Water Use for Precision
Agriculture with UAV-Based Sensors.Remote Sensing of Environment, 167, 140-152.

[10] Bukar, A. L., et al. (2020).Smart Agriculture: An Approach towards Future Farming.
Computers and Electronics in Agriculture, 174, 105507.

[11] Gomez-Castaneda, C., et al. (2019).Smart Irrigation Systems: Challenges and

Opportunities.Journal of Irrigation and Drainage Engineering, 145(1), 04018064.

[12] Zhang, Z., et al. (2018).Smart Irrigation Systems: State of the Art and Prospective
Research Areas.Computers and Electronics in Agriculture, 144, 57-66.

[13]https://view.officeapps.live.com/op/errorpage.htm?llcc=en-US&hid=ef7b58e6-62bc-4
952-92e7-604b6d473ac8

[14] Kumar & Patel, 2021

25
[15](Chen et al., 2018)

[16] (Garcia-Sanchez et al., 2019)

[17] Lee & Kim, 2020

[18] Brown & White, 2019

[19] Li & Wang, 2021

[20] Johnson & Smith, 2020

[21] Gomez et al., 2018

26
27