ML-Based What-If Analysis to Mitigate Interference for 5G NR Non-Public Networks

As industries become more digitized, they are entering a new era of connectivity based on 5G and beyond technology features. This transformation brings new challenges such as interference between public and non-public networks. In this study, we investigate the automatic configuration of non-public networks using a Machine Learning-based What-if Analysis technique to mitigate interference and enhance 5G NR networks coexistence. This proactive decision-making solution is designed for various coexistence scenarios, namely adjacent and co-channel. It provides a systematic approach for selecting the most suitable network parameters and configurations. We validate our approach through experimental analysis, with a particular focus on selecting and analyzing the TDD pattern and link robustness for 5G NR non-public network deployments. Our analysis is based on real-world experiments carried out at the 5G Industry Campus Europe in Aachen, Germany.