CELLULAR NETWORK TRAFFIC PREDICTION SISTEM USING MACHINE LEARNING ALGORITHM AND GATED RECURRENT UNIT BASED ON BTS TRAFFIC DATA

• Main Author: Maliky, Anggoro
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/87886
• Institution: Institut Teknologi Bandung

The swift advancement of internet networks in this digital age has triggered a chain reaction in the need for dependable and extensive cellular network traffic. The internet's role as a daily essential via digital devices like smartphones, streaming platforms, and cloud services has also led to a rise in cellular network traffic. The rising demand presents difficulties in overseeing cellular network resources because of the escalating complexity of cellular network traffic. To tackle these challenges, a prediction system that can reliably forecast cellular network traffic is essential to aid network administrators in making informed decisions for resource management. The reliance on outdated prediction algorithms, whichfail to address the intricacies of traffic surges, has led to a demand for a more dependable prediction system utilizing innovative algorithms. The emergence of artificial intelligence technology offers a new viewpoint on addressing the urgent demand for cellular network traffic forecasting. In this Capstone project, the execution of prediction algorithms starts with machine learning featuring decision trees, progresses to ensemble learning via Extra Gradient Boosting, and culminates in deep learning prediction methods like Gated Recurrent Unit. This method is hoped to function as a decision-making resource to assist network administrators in efficiently managing cellular network assets. The testing results demonstrate that all the prediction models achieve high accuracy indicate this system has the potential to be utilized as a decision-making resource to assist network administrators in efficiently managing cellular network asset.