Iridescence: Improving Configuration Tuning in the Presence of Confounders for 5G NSA Networks
Changhan Ge, Ajay Anil Mahimkar, Zihui Ge, Romeo Fernandez, Joseph Maniaci, Shomik Pathak, Maulik Shah · Proceedings of the ACM on Networking · 2025
Configuration tuning is one of the top network operational tasks for Cellular Service Providers (CSPs), and is typically done to either restore performance during degraded network conditions such as congestion, failure, planned upgrades, or optimize service performance through change trials. A long-standing challenge in tuning has been to associate a causal relationship between a configuration change and a service performance impact. Confounders (or, external factors) make this extremely hard. In this paper, we focus on improving configuration tuning in the presence of confounders for 5G Non-standalone (NSA) networks. We propose a new solution Iridescence that uses advanced machine learning techniques such as XGBoost or transformers to first de-confound the performance impacts, and then improve the impact classification process for configuration tuning. We thoroughly evaluate Iridescence using a very large data set collected from an operational 5G NSA network. We share our findings with network engineering and operations teams and confirm the configuration changes that have high likelihood of improving the 5G NSA performance. Our preliminary trials demonstrate that Iridescence can achieve performance improvements in operational 5G networks.