DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri

DDoS attacks represent a substantial danger to network availability and security. Because of their sophistication, traditional DDoS detection technologies sometimes struggle to effectively identify and neutralise such attacks. In this project, offer a novel method for detecting DDoS attacks using Ar...

Full description

Saved in:
Bibliographic Details
Main Author: Yusri, Muhammad Syamil
Format: Thesis
Language:English
Published: 2024
Subjects:
Online Access:https://ir.uitm.edu.my/id/eprint/95729/1/95729.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
id my-uitm-ir.95729
record_format uketd_dc
spelling my-uitm-ir.957292024-05-23T03:20:56Z DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri 2024 Yusri, Muhammad Syamil Neural networks (Computer science) DDoS attacks represent a substantial danger to network availability and security. Because of their sophistication, traditional DDoS detection technologies sometimes struggle to effectively identify and neutralise such attacks. In this project, offer a novel method for detecting DDoS attacks using Artificial Neural Networks (ANNs). To detect aberrant network traffic associated with DDoS attacks, the system uses the capability of ANNs, which are capable of learning complicated patterns and generating accurate predictions. The suggested system is divided into two stages: training and detection. A systematic strategy is used in the study framework, which includes problem identification, data gathering, preprocessing, ANN implementation, and performance assessment. The project's goal is to improve network security by identifying and categorising DDoS attacks properly. The implementation of ANNs using the Sequential model was tested using different training/testing splits (80/20, 70/30, 60/40) and epochs. The 70/30 split regularly outperformed others, with an accuracy rate of 92.29%. Detailed parameter tweaking was carried out for the 80/20, 70/30, and 60/40 splits, finding that a three-hidden-layer architecture with 256, 128, and 8 neurons produced the highest accuracy of 92.54%. Multiple ANN models were evaluated, and the best-performing model achieved 92.54% accuracy, 93.01% precision, 91.05% recall, and a 92.10% F1-score. However, the research recognises several limits, such as the time-consuming nature of ANN model training, possible scaling concerns due to hardware restrictions, and fluctuation in dataset appropriateness. To overcome these constraints, the project propose future recommendations like as using advanced techniques like parallel processing, automated methods for dynamic dataset selection, and hybrid approaches that combine ANN with other methods to increase accuracy. 2024 Thesis https://ir.uitm.edu.my/id/eprint/95729/ https://ir.uitm.edu.my/id/eprint/95729/1/95729.pdf text en public degree Universiti Teknologi MARA, Terengganu College of Computing, Informatics and Media Abdul Latif, Mohd Hanapi
institution Universiti Teknologi MARA
collection UiTM Institutional Repository
language English
advisor Abdul Latif, Mohd Hanapi
topic Neural networks (Computer science)
spellingShingle Neural networks (Computer science)
Yusri, Muhammad Syamil
DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri
description DDoS attacks represent a substantial danger to network availability and security. Because of their sophistication, traditional DDoS detection technologies sometimes struggle to effectively identify and neutralise such attacks. In this project, offer a novel method for detecting DDoS attacks using Artificial Neural Networks (ANNs). To detect aberrant network traffic associated with DDoS attacks, the system uses the capability of ANNs, which are capable of learning complicated patterns and generating accurate predictions. The suggested system is divided into two stages: training and detection. A systematic strategy is used in the study framework, which includes problem identification, data gathering, preprocessing, ANN implementation, and performance assessment. The project's goal is to improve network security by identifying and categorising DDoS attacks properly. The implementation of ANNs using the Sequential model was tested using different training/testing splits (80/20, 70/30, 60/40) and epochs. The 70/30 split regularly outperformed others, with an accuracy rate of 92.29%. Detailed parameter tweaking was carried out for the 80/20, 70/30, and 60/40 splits, finding that a three-hidden-layer architecture with 256, 128, and 8 neurons produced the highest accuracy of 92.54%. Multiple ANN models were evaluated, and the best-performing model achieved 92.54% accuracy, 93.01% precision, 91.05% recall, and a 92.10% F1-score. However, the research recognises several limits, such as the time-consuming nature of ANN model training, possible scaling concerns due to hardware restrictions, and fluctuation in dataset appropriateness. To overcome these constraints, the project propose future recommendations like as using advanced techniques like parallel processing, automated methods for dynamic dataset selection, and hybrid approaches that combine ANN with other methods to increase accuracy.
format Thesis
qualification_level Bachelor degree
author Yusri, Muhammad Syamil
author_facet Yusri, Muhammad Syamil
author_sort Yusri, Muhammad Syamil
title DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri
title_short DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri
title_full DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri
title_fullStr DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri
title_full_unstemmed DDoS attack detection system using Artificial Neural Network (ANN) / Muhammad Syamil Yusri
title_sort ddos attack detection system using artificial neural network (ann) / muhammad syamil yusri
granting_institution Universiti Teknologi MARA, Terengganu
granting_department College of Computing, Informatics and Media
publishDate 2024
url https://ir.uitm.edu.my/id/eprint/95729/1/95729.pdf
_version_ 1804889972981891072