Theses and Dissertations

Date of Award

7-1-2024

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Mathematics

First Advisor

Mrinal Kanti Roychowdhury

Second Advisor

Sanjeev Kumar

Third Advisor

Hansapani Rodrigo

Abstract

A Distributed Denial-of-Service (DDoS) attack involves overwhelming a target system's data bandwidth or computational resources, often using multiple attack systems, aiming to slow down or disable the targeted system. Detecting and mitigating DDoS attacks effectively remains challenging due to their varying characteristics. One of the promising approaches involves developing an AI based Intrusion Detection System (IDS) against cyberattacks. In this study, we aim to develop an AI based Intrusion Detection System (IDS) for DDoS threat detection using Machine Learning, Deep Learning, or hybrid techniques. Different Machine Learning (ML) and Deep Learning (DL) algorithms like Random Forest (RF), Naïve Bayes (NB), Logistic Regression (LR), K-Nearest Neighborhood (KNN), Deep Neural Network (DNN), Long Short-Term Memory (LSTM) have been used to build the AI based intrusion detection system. CIC-DDoS-2019 and CIC-IoT-2023 datasets were utilized in this work for training and testing the performance of the AI models. In this work, we also concentrated on addressing data imbalance issues, which arose from the presence of high volume of attack data compared to benign data. Four different data balancing techniques have been used to solve the data imbalance problem. The performance of ML and DL models was assessed using metrics such as accuracy, precision, recall, F1 score, balanced accuracy, and Area Under the ROC-Curve (AUC) score under four different balancing techniques. Lastly, we compared the performance of these ML and DL models with different balancing techniques to obtain a better solution.

Comments

Copyright 2024 Dipok Deb. https://proquest.com/docview/3116042070

Download

Included in

Mathematics Commons

Share

COinS