Data-Driven Intelligent Outage Management for High Shadowing Environments in 5G&B Networks

In the evolving landscape of 5G and forthcoming 6G networks, managing outages becomes increasingly complex due to higher Base Station (BS) densities and the consequent rise in outage instances. Addressing this, we introduce a sophisticated, two-tiered outage management framework that leverages artificial intelligence for enhanced efficiency and automation. Our approach features an innovative AI -based cell outage detection strategy, named Impv-XGBoost, which excels in high-shadowing conditions and with sparse training data, outperforming traditional methods. The framework's second tier employs an actor-critic reinforcement learning scheme for cell outage compensation, finely tuning compensating BS's tilt and transmit power. This method uniquely integrates outage and compensating base station's user equipment coverage, ensuring equitable service quality. By incorporating Jain's fairness index and the geometric mean in its reward mechanism, our approach achieves fair and efficient outage management, demonstrating notable improvements in user coverage during BS outages.