Measurement-Driven Early Warning of Reliability Breakdown in 5G NSA Railway Networks

TL;DR

This study benchmarks machine learning models for seconds-ahead reliability prediction in 5G NSA railway networks using real measurement data, demonstrating the feasibility of early warning for reliability breakdowns.

Contribution

It develops a measurement-driven benchmark for reliability prediction, evaluating multiple models and providing insights into operating trade-offs in 5G railway environments.

Findings

Models can predict reliability breakdowns seconds in advance.

Lightweight radio features are sufficient for effective prediction.

Benchmark offers empirical insights for sensing-assisted communication control.

Abstract

This paper presents a measurement-driven study of early warning for reliability breakdown events in 5G non-standalone (NSA) railway networks. Using 10 Hz metro-train measurement traces with serving- and neighbor-cell indicators, we benchmark six representative learning models, including CNN, LSTM, XGBoost, Anomaly Transformer, PatchTST, and TimesNet, under multiple observation windows and prediction horizons. Rather than proposing a new prediction architecture, this study develops a measurement-driven benchmark to quantify the feasibility and operating trade-offs of seconds-ahead reliability prediction in 5G NSA railway environments. Experimental results show that learning models can anticipate RLF-related reliability breakdown events seconds in advance using lightweight radio features available on commercial devices. The presented benchmark provides insights for sensing-assisted communication control and offers an empirical foundation for integrating sensing and analytics into future mobility control.