Intelligent Cellular Performance Monitoring and Fault Diagnosis Using Deep Graph Analytics

Abstract

The continuous evolution of wireless communication technologies, including WiFi, 4G, and 5G, has generated vast amounts of network data that require intelligent analysis to ensure reliable connectivity and efficient communication. Accurate prediction of network characteristics, such as connection type and signal strength, has become increasingly important for maintaining service quality, optimizing resource utilization, and improving user experience. Conventional network monitoring techniques mainly rely on visualization tools that display performance indicators such as latency, throughput, and received signal strength. Although these approaches assist in monitoring network conditions, they provide limited predictive capabilities and require significant manual interpretation. In addition, traditional methods often fail to perform effective data preprocessing, including missing value handling, categorical feature encoding, and data normalization, which can negatively affect prediction accuracy. To overcome these challenges, this work presents a Flask-based web application that integrates machine learning techniques for intelligent network signal analysis and prediction. The proposed framework employs Ridge Classifier (RC), Ridge Regressor (RR), Decision Tree Classifier (DTC), Decision Tree Regressor (DTR), and a Hybrid MLP with Catboost model to perform both classification and regression tasks. Network type identification is formulated as a classification problem, whereas signal strength estimation is treated as a regression problem. The developed system evaluates model performance using accuracy, precision, recall, F1-score, MAE, MSE, RMSE, and R² metrics, enabling comprehensive comparison of predictive models. The proposed framework offers an efficient, scalable, and user-friendly solution for intelligent network performance analysis and future wireless communication applications.