DDOS Attack Detection using Machine Learning and Deep Learning Methods
Thesis Type: Postgraduate
Institution Of The Thesis: TED University, Faculty of Engineering, Department of Software Engineering, Turkey
Approval Date: 2024
Thesis Language: Turkish
Student: Deniz Merve Gündüz
Supervisor: Emin Kuğu
Open Archive Collection: AVESIS Open Access Collection
Abstract:
Network infrastructure security and stability are seriously threatened by Distributed Denial of Service (DDoS) attacks. The demand for sophisticated detection systems that can recognize and neutralize threats instantly is rising as these attacks get more and more complex. In this thesis, the CSE-CIC-IDS2018 dataset is used to study the application of machine learning and deep learning methods to the detection of DDoS attacks. In this work, we investigate how well two deep learning models: Artificial Neural Networks (ANN) and Long Short-Term Memory (LSTM) networks along with two machine learning algorithms: Random Forest and Support Vector Machine (SVM) identify DDoS attacks. To extract useful features from the CSE-CIC-IDS2018 dataset, two feature selection techniques: information gain and correlation were used to maximize model performance. Every model completes training and evaluation based on its precision in differentiating between harmful and legal traffic. The findings show that deep learning models—LSTM in particular—perform very well at identifying the temporal patterns present in network data, which raises detection rates in situations where time is critical. On the other hand, the Random Forest and SVM models offered a compromise between detection accuracy and computational economy, delivering strong performance with designed features. A comparative study of these models reveals their advantages and disadvantages in relation to DDoS detection, offering important new information for further investigation and real-world cybersecurity applications. By providing an extensive evaluation of modern machine learning and deep learning methods for DDoS attack detection, this thesis advances the subject of cybersecurity and provides a study that may be expanded to include various types of network-based attacks. Keywords: deep learning, machine learning, ddos attack, ddos attack detection