Hidden Markov Model Approach to Predict Performance Improvement of 4G Cellular Network

The automation of performance optimization in 4G cellular networks remains a significant research challenge. Although advances in artificial intelligence, machine learning, and deep learning have been made, their application in cellular networks is still in its early stages. The evolution of automated learning algorithms in these networks requires continuous optimization to improve performance. This research aims to develop a data analysis model using a Machine Learning approach to predict 4G network Key Performance Indicators (KPIs) and enable automated assessments. The study employs Hidden Markov Models (HMM) to predict KPIs such as eRAB Drop Rate and Setup Success Rate. Data were collected from 1,600 newly deployed sites of a mobile operator in Indonesia over six months, from July 2023 to January 2024. The findings indicate that HMM models can significantly contribute to understanding and enhancing cellular network performance. For instance, the model achieved an accuracy of 99.52% and an F1 score of 99.76% in predicting the eRAB Drop Rate. The Inter-Frequency Handover Success Rate showed 96.15% accuracy and a 97.99<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>F1 score, while the RRC Connection Setup Success Rate achieved 91.35% accuracy and a 95.29% F1 score. These metrics highlight the model's effectiveness in predicting KPIs. By utilizing HMM models for high-accuracy KPI analysis, network operators can gain deeper insights into network performance and implement targeted improvements. This approach presents a promising avenue for automating and enhancing performance in 4G networks.