ENCRYPTED TRAFFIC CLASSIFICATION IN 5G NETWORKS: DESIGN, IMPLEMENTATION, AND COMPARATIVE EVALUATION OF DEEP LEARNING TECHNIQUES

This thesis analyzes the classification of encrypted 5G network traffic using both machine learning and deep learning approaches. By testing different models (Random Forest and Convolutional Neural Networks) on real 5G traffic data, the study compares their accuracy, robustness, and efficiency. Results show that both approaches are effective, with deep learning excelling on short time windows and Random Forest providing strong stability overall. The work contributes to understanding how encrypted traffic can be managed and classified for enhanced network security.