DNS DDoS Amplification Attack Detection Using Multi-Layer Perceptron Classification Algorithm

Distributed Denial-of-Service (DDoS) attacks pose a significant threat to online services, particularly those targeting Domain Name System (DNS) amplification. This study investigates the effectiveness of a Multi-Layer Perceptron (MLP) classifier for detecting DNS amplification attacks within network traffic data. The MLP employs three optimization techniques (ADAM, SGD, and LBFGS) and four activation functions (identity, logistic, Tanh, and ReLU) to optimize its performance. Feature selection utilizes Pearson correlation analysis to identify the most relevant features for attack detection. The DrDoS_ DNS dataset, derived from the CIC-DDoS2019 evaluation dataset, provides the network traffic data for this research. The results demonstrate that the MLP classifier with the ADAM optimization technique achieves the best classification accuracy for detecting DNS amplification attacks. Additionally, both Tanh and ReLU activation functions exhibit strong performance within the MLP model. These findings highlight the potential of machine learning techniques like MLPs for accurately identifying and mitigating DNS amplification DDoS attacks. Future research directions include exploring Principal Component Analysis (PCA) for feature extraction and potentially improving model performance.