Brute Force Attack Detection in Network Traffic Using Convolutional Neural Networks

Abstract

Introduction: This study presents a deep learning-based approach for detecting brute force attacks in network traffic using a Convolutional Neural Network (CNN) model. Methodology: Flow-based data from the NF-UQ-NIDS dataset was preprocessed and balanced using the Downsampling and Synthetic Minority Over-sampling Technique (SMOTE) techniques. The CNN architecture was designed to extract temporal and spatial features from the input data, enabling accurate binary classification between brute force attacks and benign traffic. Results: The model was evaluated using standard performance metrics, including accuracy, precision, recall, F1-score, and AUROC that revealed exceptional results from the CNN-SMOTE configuration achieving an accuracy of 99.82% and a recall of 99.95%. Comparative analysis against benchmark models from previous studies confirmed the superiority of the proposed approach, particularly in handling class imbalance. Conclusion: The results demonstrate that deep learning models, especially when trained with the appropriate data balancing technique, can significantly enhance intrusion detection systems. Recommendations for further improvement include exploring hybrid models and integrating explainable AI components.