TelcoAgent Offers Scalable, Explainable 5G Network Forecasting

This research introduces TelcoAgent, a groundbreaking foundation model-based framework aimed at revolutionizing proactive network management for 5G and future telecom networks. Current machine learning approaches for Key Performance Measurement (KPM) forecasting often fall short in terms of scalability and explainability, limiting their real-world utility. TelcoAgent addresses these limitations by providing accurate, scalable, and explainable forecasting of multiple KPMs across diverse network cells, crucially without requiring site-specific training. The framework is built upon three core components. First, an automated three-agent pipeline is responsible for constructing a comprehensive 3rd Generation Partnership Project (3GPP) knowledge graph directly from specification documents. This ensures that the system's understanding is grounded in industry standards. Second, a scalable time-series foundation model (TSFM)-based prediction pipeline delivers accurate, zero-shot forecasting capabilities, allowing the model to generalize across different network cells. Finally, a reasoning and explanation pipeline provides actionable, domain-grounded diagnostics, offering clear insights into network degradations and suggesting corrective actions. TelcoAgent's effectiveness was validated using a three-month, real-world 5G KPM dataset from a major U.S. network operator, covering 200 cells. The system demonstrated high forecasting accuracy for all seven KPMs considered per cell. Beyond accuracy, its ability to deliver explainable insights and actionable instructions makes it a powerful tool for addressing network degradations efficiently and proactively.