School Of Computer Science Engineering and

Information Systems-(SCORE)

Title: Rainfall Prediction System

Done by
20MIS0221 - V. Varun
20MIS0289 - G. Sreehasa
20MIS0293 - K. L. V. Pramod

For the course

Course Code:
SWE2022

Course Name:

Software Engineering Process, Tools and Methods

Semester:
Fall-Semester-2023-24

1
 Content

Sno. Topic Page Number

1 ABSTRACT 3
2 INTRODUCTION 4,5
3 CHOICE OF MODEL 6,7,8
4 ARCHITECTURE 9,10
5 UML CLASS DIAGRAM 11,12,13,14
6 DATA BASE DESIGN 15,16
7 TOOLS USED 17,18,19
8 TEST CASES 20,21
9 CONCLUSION 22
10 REFERENCES 23,24

2
 1. ABSTRACT
Climate is significant factor in human life. Changing climatic conditions affect humans’ day to day
activities significantly. There are many phases of climate which are summer, winter, rainfall and
rainfall. Out of which we are focusing on rainfall climatic condition. Rainfall is vital component of
human life and is a critical resource for various activities. Accurately predicting the rainfall shall
help in many aspects such as agriculture which helps in harvest. Farmers can plan their harvest
accordingly, in water resource management; authorities can plan water release schedules based on
anticipated rainfall thereby preventing overflow and depletion of resources, disaster preparedness,
flood management etc. At the same time the inaccuracy in rainfall prediction can lead to crop
damage, inefficiency in flood management and poor disaster preparedness. There is traditional
weather forecast systems which also give the prediction but they do it on a broader scale, they don’t
specifically focus on aspects of rainfall and give predictions for greater span of time. Whereas, the
rainfall prediction system focusses on prediction in every small region, and for every one- two hours,
its’ also details on reports of intensity and duration of rainfall.

While there are many rainfall prediction systems but they vary in the accuracy. Traditional
prediction systems are not satisfactorily accurate hence, we have proposed the rainfall prediction
with Artificial Neural Network (ANN).

In this project, Artificial Neural Network (ANN) such as Feed Forward Neural Network
(FFNN) model is built for rainfall prediction task. For prediction, artificial neural networks (ANN)
are a valuable and appealing soft computing technique. ANN, including the FFNN uses self-
adaptive technique that means the model is capable of learning from historical data to capture
functional correlations between data and can also facilitate the forecast of future data. The focus
of this model is specifically relevant in context of eater resource management where accurate
rainfall predictions play a vital role. The precision in prediction shall help in planning the
utilisation of water resources. To determine the accuracy of the model performance metrics like
confusion matrix and the root mean square error (RMSE) are used. The confusion matrix predicts
the ability of model to correctly predict the classes of rainfall and RMSE quantifies the summed
up prediction error.

From the results, we understand that ANN has secured greater accuracy among rest. The
successful application of the FFNN model in rainfall prediction highlights the potential of artificial
neural networks (ANN) as reliable tools in soft computing for meteorological predictions and, in
broader aspect aid in addressing challenges related to water resource management.
3
 2.INTRODUCTION

Rainfall is fundamental aspect of our natural environment. It lays significant impacts on flood
management, water resource management, disaster preparedness and urban planning. Timely and
accurate prediction of rainfall is very crucial to make informed decisions in these domains. We have
traditional weather forecast systems which make the predictions but unfortunately they lack in
precision and localization required for effective planning. They give prediction on broader aspects
such as broader scale of regions, include other aspects other than rainfall and don’t detail the forecast
for every min. They are often not region-specific.

The proposed system in this projects employs Feed Forward Neural Network which is a variant of
Artificial Neural network (ANN). Artificial neural networks are very capable for computing
prediction tasks. The underlying mechanism of ANN has self adaptive learning process. The model
learns from historical data, identifies significant correlations within the data, and makes use of this
knowledge to make accurate predictions based on current data. From the results obtained, we
understood that ANN has secured the highest accuracy among the rest for rainfall prediction.

The NEED of rainfall prediction system:

Many crucial activities like agriculture solely rely upon rainfall to plan their harvest, the release
schedule of dams and reservoirs is also relied on predictions of rainfall, the disaster preparedness
program has iactions planned based on the rainfall predictions.There are many ore needs as such
which are fufilled by rainfall prediction system. With the need of rainfall prediction system, we also
lay down focus on the accuracy of rainfall prediction system. Having a rainfall prediction system
alone is not sufficient , it is required that system has high accuracy so that the mentioned needs can
be fulfilled properly. Else errors in accuracy can cause crop damage, lead to depletion of water
resources due to unwanted release of water and possible chance of disaster like floods, drought. To
these problems there is strong need for accurate and robust rainfall prediction system.

The REQUIRMENETS for rainfall prediction system

4
Functional requirements:
i. Login - The authentication is must to prevent use from unauthorized entity
ii. Data capture – The system should be capable to capture data from various sources
iii. Data integration – The system should be capable of integrating historic rainfall data to
identify learning patters
iv. Region selection feature – The user should be able to select the region he is looking
prediction for
v. Time period selection feature – The user should be able to select the time period the
prediction is requested for
vi. iv. artificial neural network training - The system should have a functionality to train the
feed-forward neural network (FFNN) using historical data. It must continuously update the
neural network to adapt to changing pattern
vii. Predict feature- The system should accurately predict the rainfall
viii. Post feedback feature – User should be able to post feedback and report any complaints

Non-functional requirements
i. Performance – The should provide predictions on time and should respond to changes
quickly
ii. Reliability – The system should provide reliable predictions
iii. Scalability – The system should be able to manage varying size and load of requests.
iv. Security – The system should offer security to user details through encryption mechanisms.
v. Reusability – The system should be built such that its components can be reused
vi. Extensibility – The should be capable of working with addition of new features or
modification of existing features as per the user needs.
vii. Availability – The system should be available as and when the suer need its
viii. Usability – The system should offer friendly UI to its users.

Environmental requirements
i. Data source compatibility – The system should be compatible of different formats of sources
of data
ii. Proper network connectivity – The system is required to have proper network connectivity.
iii. Storage- Requirement of total RAM capacity of 64 GB for overall performance.
iv. Optimize power consumption – The power consumption should be optimized without effect
on system performance

5
3.CHOICE OF MODEL : ITERATIVE INCREMENTAL MODEL

JUSTIFICATION
1. Nature of Predictive Modeling:
• Rainfall prediction deals with with complex algorithms, statistical models, and
evolving data patterns. An iterative approach allows to refine these models over
successive cycles and accommodate changes.

2. Uncertainty in Requirements:
• This system has evolving requirements which are well accommodated by incremental
and iterative model.

3. Continuous Improvement:
• Each iteration lays focus on enhancing certain aspects like UI, accuracy etc, This
model gives platform for such continuous improvements

4. Adaptation to Stakeholder Feedback:

• This model allows for continuous feedback and reviews from stakeholders to enhance
the performance of the system.

5. Risk Management:
• The Rainfall Prediction System may encounter risks related to changes in data
patterns, unexpected events, or evolving user requirements. The model provides risk
management through repeated cycles helps address and mitigate these uncertainties
6
 6. Early Delivery of Value:
• The Incremental aspect of this model provides early built and delivery of valuable
features where stakeholders can benefit and from certain functionalities even before
the entire system is complete and can convey what improvements they are expecting.

ITERATIONS OF OUR SYSTEM

Iteration 1:
1. Requirements Gathering: Obtain functional, non functional requirments
2. Initial Design and Data Collection: The system architecture is prepared and data is collected
3. Basic Neural Network Implementation: The feed forward neural network are implemented
4. User Interface Prototyping: Prototypes of sample UI are delivered
5. Evaluation and Feedback: Feedback are recorded.

Iteration 2.
1. Advanced Neural Network Features: Implemented ffnn is enhanced with more advanced features
2. Expanded User Functionality: From the feedback recorded in iteration 1, new func. Are added
3. Integration of Real-Time Data: In iteration1 we dumped data sources, now we integrate with real
time data capture system
4. Iterative Testing and Debugging: Interactively test and debug the system

Iteration 3:
1. Optimization and Performance: Improve the speed of prediction
2. User Feedback Incorporation: Incorporate and demonstrate implemented feedback of users
3. Documentation and Training: Document and train personnels.
4. Final Evaluation and Deployment: Deploy the final system

Ongoing Iterations:
1. Continuous Monitoring and Improvements
2. Adapt to Environmental Changes.

7
MODULES:

1. Login module – Authenticate the suer to prevent use from unauthorized entity

2. Selection of choice module – Allows users to select choices like region where they are
looing prediction for and time period until which they want the prediction etc.,

3. Data collection module- Collects data from various sources to predict rainfall

4. Data preprocessing module- Cleaning of noisy and redundant data

5. Feature extraction module – Extracts features that can be input to predict rainfall like
temperature, humidity and wind speed

6. ANN training module- Optimism the neural network’s parameters for accurate rainfall
prediction

7. Real time data integration – Allow the system to integrate real time data and make
predictions

8. Prediction module – Based the data obtained from various modules, it predicts the rainfall

9. UI module – responsible for delivery of user friendly UI

10. Evaluation and Report module – Responsible to evaluate the accuracy of prediction and
generate reports.

11. Security module – Protects the privacy of user details,

12. Feedback module – Records the suggestion and complaints.

8
 4.Architecture

Architecture Diagram for Rainfall Prediction System PAST WEATHER

DATABASE

login

Register

SELECT MODULE DATA COLLECTION MODULE

USER DATABASE

DATA PREPROCESSING MODULE

FEATURE EXTRACTION MODULE

ANN TRAINING MODULE PREDICTION MODULE

REAL TIME DATA
INTEGRATON MODULE

USER DATABASE

SECURITY MODULE REPORT GENERATION EVALUATION MODULE

MODULE

DOCUMENTATION AND MAINTAINENCE MODULE FEEDBACK MODULE

TRAINING MODULE

9
Explanation
1. Login module -:
The user logs in if he is already registered else gets registered if he hasn’t registered
2. Select module:-
The user selects the factor he seeks for rainfall prediction like the region where he want the
prediction of rainfall, the time period for which the rainfall occurs and the intensity of
rainfall
3. Data collection module:-
The data is gathered from historical meteorological data from weather stations, satellites
and other sources.
4. Data preprocessing module-:
This module handles missing data, removes outliers and scales features.
5. Feature extraction module-:
This module identifies and extract relevant features from the preprocessed data that will
determine the rainfall like the wind, speed, and temperature
6. Artificial Neural Network (ANN) Training Module:
Initializes the FFNN model and trains the model using historical data to capture patterns
and correlations.
7. Real time data integration module-:
The model has been trained with past data, now the real time weather consitions are
captured and duped into the model for prediction purpose
8. Prediction model
Predicts the rainfall based on the data given as input
9. Evaluation module
This modules evaluates the accuracy of the prediction using confusion matrix and uses
RMSE
10. Report generation module
This module generates reports of rainfall prediction so that suer can view them
11. Security module
This module secures the suer database and reports using security mechanisms like encryption
12. Documentation and training module
Documents and trains the personnel
13. Feedback and maintenance module
Checks for repairs and collects feedback from key stakeholders and users

10
 5. UML CLASS diagram

Here, the complete classes involved in the project should be represented as

UML-CLASS diagram and its methods along with all attributes

11
RELATIONSHIP BETWEEN CLASESS
1. Association
2. Realizations
3. Inheritance
4. Dependency
5. Aggregation
6. Composition

1. Association
Represents a relationship between two or more classes, indicating that instances of one
class are somehow connected to instances of another class
Example: If class A is associated with class B, it could mean instances of A use instances
of B in some way.

2. Realization (Interface):
Realization is a relationship between a class (or a component) and an interface, indicating
that the class implements the operations defined by the interface.
Example: Class A realizes Interface B, indicating that A provides the concrete
implementation of the operations declared in B

3. Inheritance
Inheritance represents an "is-a" relationship, where one class is a more specialized version
of another class.
Example: If class A inherits from class B, it means A is a more specialized version of B.

4. Dependency
Dependency indicates that one class depends on another class, often in terms of method
calls or parameter types
If class A depends on class B, it means changes in B may affect A.

12
 5. Aggregation
Aggregation represents a "has-a" or "part-of" relationship, where one class contains or is
composed of another class
Example: If class A aggregates class B, it means instances of A contain instances of B.

6. Composition
Composition is a stronger form of aggregation, where the lifetime of the part is tied to the
lifetime of the whole
Example: If class A composes class B, it means instances of A own and manage the
instances of B.

RELATIONSHIPS IN RAINFALL PREDICTIONS SYSTEM

Association:
Between RainfallPredictionSystem and Prediction:
RainfallPredictionSystem uses Prediction for making rainfall predictions.
Between RainfallPredictionSystem and DataCollection:
RainfallPredictionSystem uses DataCollection to gather historical data.
Between RainfallPredictionSystem and DataPreprocessing:
RainfallPredictionSystem uses DataPreprocessing for cleaning and normalizing data.
Between RainfallPredictionSystem and FeatureExtraction:
RainfallPredictionSystem uses FeatureExtraction for extracting relevant features.
Between RainfallPredictionSystem and Evaluation:
RainfallPredictionSystem uses Evaluation for assessing prediction accuracy.
Between RainfallPredictionSystem and Report:
RainfallPredictionSystem uses Report for generating reports.
Between RainfallPredictionSystem and Feedback:
RainfallPredictionSystem may have a feedback mechanism using Feedback.
Between RainfallPredictionSystem and SecurityComponent:
RainfallPredictionSystem collaborates with SecurityComponent for ensuring security.

Dependency:
Between Prediction and FeatureExtraction:
Prediction may depend on FeatureExtraction for extracting features from data.

13
Between Report and Evaluation:
Report may depend on Evaluation for including assessment results in reports.
Between SecurityComponent and RainfallPredictionSystem:
SecurityComponent ensures security for RainfallPredictionSystem.

Aggregation:
RainfallPredictionSystem may contain Feedback as part of its functionality.

14
 6.DATA BASE DESIGN as graphical representation

Complete data base design of the project to be included here in the graphical
form with explanations and its relations

The proposed database design encompasses key entities and relationships for a
comprehensive rainfall forecasting system. The system begins with dataset acquisition,
storing weather-related data with timestamps. The preprocessing stage involves feature
extraction and correlation matrix calculation, resulting in preprocessed data linked to the
original dataset. Two forecast models are then implemented: an LSTM-based model
generating 8-hour predictions and 1-step forecasts, and an XGBoost-based model providing
1-hour predictions. Model evaluations capture actual rainfall data, predicted rainfall data
from both models, and are timestamped. The design ensures data integrity through foreign
key relationships, tracks temporal aspects with timestamps, and incorporates features
extracted during preprocessing. This structured approach enables efficient analysis and
evaluation of rainfall predictions aligned with a 1-hour interval.

15
 Entities:

Dataset:

Attributes: DatasetID, Timestamp, Weather Features, Rainfall Data

PreprocessedData:

Attributes: PreprocessID, DatasetID (foreign key), Extracted Features, Correlation Matrix

LSTMForecast:

Attributes: LSTMForecastID, PreprocessID (foreign key), 8-Hour Prediction Data, 1-Step Forecast
Data
XGBoostForecast:

Attributes: XGBoostForecastID, PreprocessID (foreign key), 1-Hour Prediction Data

ModelEvaluation:

Attributes: EvaluationID, Timestamp, Actual Rainfall Data, Predicted Rainfall Data (LSTM), Predicted
Rainfall Data (XGBoost)
Relationships:

1. Dataset -> PreprocessedData:

One-to-One relationship: One dataset can be preprocessed once.

2. PreprocessedData -> LSTMForecast:

One-to-One relationship: One preprocessing result can have one LSTM forecast.
3. PreprocessedData -> XGBoostForecast:

One-to-One relationship: One preprocessing result can have one XGBoost forecast.
4. Dataset -> ModelEvaluation:

One-to-One relationship: One dataset can have one model evaluation

16
 7.Tools Used

VISUAL STUDIO CODE

Visual Studio Code (VS Code) is a highly popular, free, and open-source source code editor
developed by Microsoft. Key features of VS Code include cross-platform support for Windows,
macOS, and Linux, a clean and intuitive user interface, and extensive customization through a
wide array of extensions available in the VS Code Marketplace. It provides built-in support for
numerous programming languages, with intelligent code completion (IntelliSense) based on
context.

Integrated Git control allows version control directly within the editor, while debugging tools
support multiple languages for a seamless debugging experience. The editor includes an integrated
terminal, task automation for running build tools, and Live Share extension for collaborative
coding in real-time. Themes and customization options enable users to personalize the appearance
of the editor. The community-driven development model ensures regular updates and ongoing
improvements, making VS Code a versatile and continuously evolving tool. With integrations into
various services and platforms, support for code snippets, multi-cursor editing, and a focus on user
productivity, Visual Studio Code has become a preferred choice for developers across different
domains.

SORTSITE – Testing tool

18
SortSite is a website testing tool designed to thoroughly assess websites for various issues such as
accessibility, broken links, spelling errors, and adherence to web standards. It serves as a valuable
resource for website owners and developers by ensuring compliance with WCAG and Section 508
guidelines. The tool covers a wide range of file types, including identifying flashing GIFs and
untagged PDFs. Additionally, it conducts checks for broken links, spelling errors, and
compatibility issues across different browsers. SortSite goes beyond these functionalities to assess
search engine optimization based on Google and Bing webmaster guidelines. It also validates
HTML and CSS against web standards and assesses usability according to Usability.gov
guidelines. This comprehensive testing tool provides automated testing, detailed reporting, and
actionable suggestions for issue resolution. It is a widely utilized resource among web developers,
designers, and quality assurance professionals, contributing to the creation of fully functional,
standards-compliant, and accessible websites. SortSite can be obtained either as a standalone
application or as part of a suite of web testing tools.

JUPYTER NOTEBOOK

Jupyter Notebook is an open-source interactive web application that allows users to create and
share documents containing live code, equations, visualizations, and narrative text. It supports
various programming languages, with Python being the most widely used. Jupyter Notebooks
provide an interactive and exploratory computing environment, making it popular among data
scientists, researchers, and educators. The user-friendly interface encourages the creation of
documents that seamlessly combine code execution, visualizations, and explanatory text, fostering
collaborative and reproducible research.

19
 8.Test Cases

1. User Authentication:
Test Case 1: Successful Login
Input: Valid username and password
Expected Output: User gains access to the system.

Test Case 2: Failed Login

Input: Invalid username or password
Expected Output: User receives an authentication error.

2. Data Collection:
Test Case 3: Historical Data Retrieval
Input: Request to collect historical rainfall data
Expected Output: System retrieves and stores historical data.

Test Case 4: Real-time Data Integration

Input: Integration with real-time weather data
Expected Output: System updates with the latest weather information.

3. Prediction Engine:
Test Case 5: Rainfall Prediction Accuracy
Input: Historical data and current weather conditions
Expected Output: Predicted rainfall aligns with actual observations.

4. Data Preprocessing:
Test Case 6: Data Cleaning
Input: Raw, unprocessed data
Expected Output: Cleaned and normalized data for accurate predictions.

5. Feature Extraction:
Test Case 7: Relevant Feature Extraction
Input: Processed data
Expected Output: Extracted features that contribute to accurate predictions.
20
6. User Interface:
Test Case 8: User Input Handling
Input: User preferences and settings
Expected Output: System adjusts settings based on user input

Test Case 9: Results Display

Input: Rainfall predictions
Expected Output: Clear and understandable display of prediction results.

7. Evaluation and Reporting:

Test Case 10: Prediction Evaluation
Input: Actual rainfall data and predicted values
Expected Output: Evaluation metrics (e.g., confusion matrix, RMSE) are calculated accurately.

Test Case 11: Report Generation

Input: Evaluation results
Expected Output: System generates detailed reports on prediction performance.

8. Feedback Mechanism:
Test Case 12: User Feedback Submission
Input: User provides feedback on prediction accuracy
Expected Output: Feedback is recorded and may influence system improvements.

9. Security:
Test Case 13: Unauthorized Access Attempt
Input: Attempted access without proper authentication
Expected Output: Access denied; security measures prevent unauthorized entry.

10. Environmental Adaptation:

Test Case 14: System Adaptation to Changes
Input: Significant environmental changes (e.g., climate shift)
Expected Output: System adapts its algorithms to new conditions.

21
 9.CONCLUSION
Rainfall prediction, a task of high significance plays crucial aspect in environmental
management, agriculture, disaster preparedness, and urban development. The implementation of
a robust and accurate Rainfall Prediction System leveraging advanced technology and data
analysis has aided in addressing the issues experienced with traditional weather forecasts. As we
conclude the development and evaluation of our Rainfall Prediction System, several key insights
emerge.
Harnessing the Power of Data Mining:
The significance of data mining techniques in predicting rainfall is very crucial. Our system relies
on extraction of hidden knowledge from historical weather data, thereby enabling a more nuanced
understanding of precipitation patterns. With the sue of data mining not only the accuracy of our
system has improved but also decision makers are provided with actionable insights .
Continuous Optimization and Integration:
In the past decade, a concerted effort by researchers has been witnessed in optimizing and
integrating data mining techniques for rainfall prediction. This ongoing refinement has resulted
in a diverse array of models and methodologies available today. However, a critical observation
is the absence of a consolidated literature review and systematic mapping study, a gap that our
system aims to address. This comprehensive analysis is vital for capturing current challenges,
proposed solutions, and emerging trends in the field.
Performance Evaluation:
The efficiency of our Rainfall Prediction System was assessed by the accuracy measure taking
the help of confusion matrix and RMSE. Comparisons have been made between predicted results
and observed results which has been considered for knowing the accuracy of our system.
Towards Informed Decision-Making:
The ultimate goal of our Rainfall Prediction System is to provide accurate and timely predictions,
to assist stakeholders in agriculture, water resource management, disaster preparedness, and
urban planning.
Future Directions:
As we conclude this phase, we understand that the journey of improving rainfall prediction is
continuous. In the future direction, we attempt to explore emerging data sources, integration of
cutting-edge machine learning algorithms, and a deeper focus on real-time adaptability to
environmental changes.

22
 10. REFERENCES

[1] https://journalofbigdata.springeropen.com/counter/pdf/10.1186/s40537-021-
00545-4.pdf
[2] https://www.sciencedirect.com/science/article/pii/S2090447920302069
[3]https://www.researchgate.net/publication/333223383_Rainfall_Prediction_U
sing_Machine_Learning
[4] https://www.mdpi.com/1424-8220/22/9/3504
[5]https://www.researchgate.net/publication/339593545_Prediction_of_Precipit
ation_Based_on_Recurrent_Neural_Networks_in_Jingdezhen_Jiangxi_Province
_China
[6] https://www.researchgate.net/publication/349707159_Prediction_of_Rainfall
_using_Machine_Learning_Algorithms_for_Different_Districts_of_Meghalaya
[7] https://ieeexplore.ieee.org/document/9702409
[8] https://ieeexplore.ieee.org/document/9155896
[9] https://arxiv.org/abs/1910.13827
[10] https://link.springer.com/content/10.1038/s41598-019-56452-5
[11] https://ieeexplore.ieee.org/abstract/document/8899704/
[12] https://iwaponline.com/jh/article-abstract/22/3/541/73683
[13] https://ieeexplore.ieee.org/abstract/document/8468083/
[14] https://www.degruyter.com/document/doi/10.1515/geo-2020-0276
[15]https://ieeexplore.ieee.org.egateway.vit.ac.in/stamp/stamp.jsp?tp=&arnumb
er=9742746
[16] https://ieeexplore.ieee.org.egateway.vit.ac.in/stamp/stamp.jsp?tp=&arnumb
er=9579472
[17] https://ieeexplore.ieee.org/document/8745763
[18] https://ieeexplore.ieee.org/document/10064510
[19] https://ieeexplore.ieee.org/document/9752220
[20] https://ieeexplore.ieee.org/document/9697017
[21] https://link.springer.com/article/10.1007/s42979-021-00592-x#Abs1
23
[22] https://ieeexplore.ieee.org/document/7279650
[23] https://ieeexplore.ieee.org/document/9768602
[24] https://ieeexplore.ieee.org/document/9702409
[25]https://www.researchgate.net/publication/357448842_A_STUDY_ON_RAI
NFALL_PREDICTION_TECHNIQUES
[26] https://ieeexplore.ieee.org/document/8597421
[27] https://www.researchgate.net/publication/343151449_Rainfall_Forecasting
_Using_Neural_Networks
[28] https://www.sciencedirect.com/science/article/pii/S1877050915020578
[29] https://ijrti.org/papers/IJRTI2010001.pdf
[30] https://www.ijrte.org/wp-content/uploads/papers/v10i4/D66111110421.pdf
[31] https://arxiv.org/pdf/2010.11787.pdf
[32]https://www.researchgate.net/publication/222489987_Forecasting_With_Ar
tificial_Neural_Networks_The_State_of_the_Art
[33] Reference: https://ieeexplore.ieee.org/abstract/document/9074233
[34]https://www.ingentaconnect.com/contentone/asp/jctn/2019/00000016/0000
0008/art00038
[35] https://www.sciencedirect.com/science/article/abs/pii/S0957417417303457
[36] https://www.sciencedirect.com/science/article/abs/pii/S0022169420302493
[37] https://www.sciencedirect.com/science/article/pii/S266682702100102X
[38] https://link.springer.com/chapter/10.1007/978-3-030-92245-0_4
[39] https://link.springer.com/content/pdf/10.1038/s41598-020-67228-7.pdf
[40] https://ieeexplore.ieee.org/abstract/document/9154184

24