Paper Details

Name: Rajni, Daljit Kaur, Inderdeep Kaur, Parminder Kaur & Harmandar Kaur
Country: India

Year: 2026
Volume: Volume No: 13, January, Year: 2026 (Special Issue)
Page Number: 16-27
DOI: https://doi.org/10.5281/zenodo.18933964

ABSTRACT
As DDoS attacks get increasingly sophisticated, traditional detection approaches fail to keep up with the changing threat landscape. Machine learning provides powerful capabilities for detecting and mitigating assaults in real time. This review paper investigates various machine learning algorithms used to detect DDoS attacks, categorizing them as supervised,
unsupervised, and deep learning approaches. Supervised learning algorithms, such as Support Vector Machines (SVM) and Decision Trees, have been widely utilized to categorize attack patterns, although unsupervised learning techniques, such as clustering, provide advantages in detecting novel assaults without the need for labeled data. Deep learning models, notably Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), have shown exceptional performance in large-scale, dynamic assault scenarios. This review also examines the role of datasets, named KDDCup99 and CICIDS, which are used to train these models, and their success is evaluated using important performance indicators such as
accuracy, precision, and recall. This study examines recent breakthroughs, datasets, and performance indicators in order to guide future research and improve the resilience of cybersecurity defenses against DDoS attacks.