5G NETWORK TRAFFIC CLASSIFICATION SYSTEM USING STACKING MACHINE LEARNING BASED ON NETWORK MONITORING TOOLS DATA

• Main Author: Rugayyah Alhaddad, Syamira
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/82147
• Institution: Institut Teknologi Bandung

The 5G network requires a traffic management system capable of accurately classifying various types of application and service traffic. Protocol behavior- based classification systems often fail to handle this complexity, resulting in decreased network performance and a subpar user experience. To address this issue, the author designed a 5G network traffic classification system using stacking machine learning to improve the accuracy and efficiency of traffic classification. In this context, the machine learning method employed is stacking machine learning, which includes k-Nearest Neighbor (kNN) and Support Vector Machine (SVM), as well as a stacking model that incorporates kNN, SVM, and Random Forest. These stacking models will be trained using data obtained from network monitoring tools to create a comprehensive 5G traffic classification system, which will be displayed on a website through an interactive dashboard. On this website, users can choose whether they only want to see visualizations regarding data traffic classification or use the website as a tool to view 5G traffic classification results using the user's network traffic data which is uploaded to the website. The system's classification results will include traffic types such as video, game, and voice. Testing conducted on each stacking machine learning model demonstrated an accuracy of 99.9% for each model, indicating that this system effectively optimizes traffic grouping based on its characteristics.