Revolutionizing Network Safety: Convolutional Neural Networks for Traffic Anomaly Detection
Jatin Sharma, Kanwarpartap Singh Gill, Deepak Upadhyay, Swati Devliyal · 2024
In the digital age, cybersecurity presents a significant challenge across various sectors and is closely linked to Sustainable Development Goals (SDGs), such as infrastructure and industrial innovation (SDG 9) and peace, justice, and strong institutions (SDG 16). Traditional anomaly detection methods are often inadequate as cyber threats grow more sophisticated and frequent. This article introduces a novel approach to anomaly detection in network traffic using Convolutional Neural Networks (CNNs). We trained and evaluated our model with the diverse set of normal and abnormal network events from the CICIDS2017 dataset, achieving an impressive accuracy of 93.64%, which surpasses that of conventional techniques. This advanced method not only enhances cybersecurity but also fosters innovation aligned with SDG 9 and supports resilient infrastructure. Additionally, it strengthens institutional frameworks by effectively identifying and mitigating cyber threats, thus advancing SDG 16. The study provides detailed performance metrics, including accuracy, recall, F1 score, and confusion matrices, demonstrating the importance of integrating advanced machine learning techniques—particularly deep learning—into cybersecurity solutions. The ability of our CNN-based model to accurately detect network anomalies highlights its potential to transform cybersecurity practices. This research advocates for continued exploration and application of deep learning methods to address the growing cybersecurity challenges, thereby supporting key SDGs and ensuring the stability and security of digital infrastructure.