Year: 2025
Type: article
Abstract: Traffic control optimization is a challenging task for various traffic centers around the world and the majority of existing approaches focus only on developing adaptive methods for normal (recurrent) traffic conditions. Artificial Intelligence (AI)-based methods have been widely adopted to predict network traffic, though with low complexity and high efficiency. This study proposes a deep learning-based intrusion detection system (IDS) using a Convolutional Neural Network (CNN) for network traffic analysis on the UNSW-NB15 dataset. The methodology involves comprehensive data preprocessing, including handling missing values, feature encoding, and addressing class imbalance, followed by feature selection using Mutual Information (MI) to enhance classification efficiency. Experimental results demonstrate that the CNN model achieves 99% accuracy, 99.03% precision, 99.86% recall, and a 99% F1-score, outperforming traditional machine learning models such as Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Logistic Regression. The ROC curve (AUC = 1.00) and confusion matrix further validate its robustness in distinguishing between normal and malicious traffic. However, early-stage overfitting is observed, necessitating further optimization. This research highlights the effectiveness of deep learning for network security, contributing to the development of scalable, real-time AI-driven security frameworks that enhance cyber threat detection and mitigation in dynamic network environments.
Institutions: Jiwaji University, Madhav Institute of Technology & Science
Language: English
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W4409395059