Abstract

DDoS attacks, which stand for Distributed Denial of Service attacks, play a significant role in impacting the reliability and availability of online services and networks. Since no system is completely immune to cybersecurity threats, which evolve daily with new techniques, studying this topic is crucial for exploring machine learning methods that can effectively detect DDoS attacks. An approach has been utilized to conduct Exploratory Data Analysis (EDA) to identify patterns suggesting the presence of DDoS attacks. Multiple machine learning models have been employed, such as Random Forest, K-Nearest Neighbors (KNN), XGBoost, and Logistic Regression. These models have undergone training and testing to identify abnormal network activity linked to DDoS attacks. Performance analysis measures, such as accuracy, recall, F1- score, and precision, are used to assess the efficiency of each model. The ML-based solution has demonstrated excellent performance in detecting DDoS attacks, as evidenced by the accurately labeled network traffic examples that determine whether they are legitimate or malicious, resulting in a calculated accuracy from the test results. Moreover, among the models used, the Random Forest and XGBoost models show exceptional accuracy, recall, and F1-score measurements, with an accuracy rate over 99%. On the other hand, while KNN shows praiseworthy performance, Logistic Regression yields somewhat lower accuracy and recall ratings.

Publication Date

5-21-2024

Document Type

Thesis

Student Type

Graduate

Degree Name

Professional Studies (MS)

Department, Program, or Center

Graduate Programs & Research (Dubai)

Advisor

Sanjay Modak

Advisor/Committee Member

Ehsan Warriach

Campus

RIT Dubai

Download

Share

COinS