Measurement-Driven Early Warning for Resilient 5G NSA Railway Networks

Po-Heng Chou; Da-Chih Lin; Hung-Yu Wei; Walid Saad; Yu Tsao

cs.NINov 12, 2025

Measurement-Driven Early Warning for Resilient 5G NSA Railway Networks

Authors: Po-Heng Chou, Da-Chih Lin, Hung-Yu Wei, Walid Saad, Yu Tsao

Organizations: Research Center for Information Technology Innovation (CITI), Academia Sinica (AS), Taipei 11529, Taiwan · Bradley Department of Electrical and Computer Engineering (ECE), Virginia Tech (VT), Alexandria, VA 22305, USA · Department of Electrical Engineering, National Taiwan University (NTU), Taipei 10617, Taiwan

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 radio link failure (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.

Figures & tables

Fig. 1 : Downlink packet loss distribution by event types and occurrence frequency.

Fig. 2 : System overview of the proposed measurement-driven early-warning framework.

Panel of Figure 3 — Fig. 3 : Time-domain results of CNN, XGBoost, and TimesNet with $T_{s}=3$ s and $I_{s}=0.1$ s. Each column corresponds to a ground-truth breakdown event, and red markers indicate prediction points that correctly anticipate the corresponding event within the specified prediction horizon.

Model	$T_{p}$	Accuracy	Precision	Recall	F1
CNN [ 13 ]	1 s	0.9806	0.2705	0.8116	0.4058
	2 s	0.9935	0.7456	0.9130	0.8208
	3 s	0.9873	0.6968	0.8551	0.7679
LSTM [ 14 ]	1 s	0.9782	0.2275	0.6957	0.3429
	2 s	0.975	0.3870	0.9058	0.5423
	3 s	0.9781	0.5374	0.7633	0.6307

TABLE I : Performance comparison with observation window

T_{s}=3

s and sampling rate

I_{s}=0.1

s (10 Hz).

Index	CNN [ 13 ]	XGBoost [ 15 ]	TimesNet [ 18 ]
Any one point	$\frac{23}{23}=100\%$	$\frac{23}{23}=100\%$	$\frac{23}{23}=100\%$
Any two points	$\frac{23}{23}=100\%$	$\frac{22}{23}=95.7\%$	$\frac{23}{23}=100\%$
Any three points	$\frac{21}{23}=91.3\%$	$\frac{20}{23}=87\%$	$\frac{23}{23}=100\%$
Two consecutive points	$\frac{17}{23}=73.9\%$	$\frac{15}{23}=65.2\%$	$\frac{16}{23}=69.6\%$
Three consecutive points	$\frac{3}{23}=13\%$	$\frac{3}{23}=13\%$	$\frac{5}{23}=21.7\%$

TABLE II : Performance evaluation of Fig. 3 .

Wireless Communications

Wireless communications studies how information is sent over radio links and how networks manage signals, spectrum, and resources reliably.

See all 399 papers

Improving 5G AI-RAN MCS Selection by Predicting Retransmissions
Sep 8, 2026Tamerlan Aghayev, Maxime Elkael, Michele Polese +5Wireless Communications
Institute for Intelligent Networked Systems, Northeastern University, Boston, MA, U.S.A. · SoftBank Research Institute of Advanced Technology, Tokyo, Japan
RIDE: An Open Dataset and Benchmark for Train Delay Prediction
Jun 3, 2026Clément Elliker, Mathis Le Bail, Clément Mantoux +2Intelligent Transportation Systems Time Series Forecasting
LIX, École Polytechnique, IP Paris, France · e.SNCF Solutions, France
ML-Based Real-Time Downlink Performance Prediction in Standalone 5G NR Using Smartphones
Mar 19, 2026Md Mahfuzur Rahman, Jareen Shuva, Nishith Tripathi +2Wireless Communications

Sep 8, 2026J/K move · Enter open · S save

Improving 5G AI-RAN MCS Selection by Predicting Retransmissions

Tamerlan Aghayev, Maxime Elkael, Michele Polese, Reshma Prasad+4

Institute for Intelligent Networked Systems, Northeastern University, Boston, MA, U.S.A. · SoftBank Research Institute of Advanced Technology, Tokyo, Japan

Wireless Communications

Open paper AskPDF arXiv

Link Adaptation (LA) in 5G NR is inherently reactive, relying on channel measurements and HARQ feedback that may become quickly obsolete when the channel changes quickly. This data is also noisy, making it hard to track accurately, and has to be fed to real-time controllers with feedback-loop effects which are hard to troubleshoot. This explains why most practical deployments select simple but robust algorithms, which accept that the lag can leave the scheduler operating at overly aggressive or unnecessarily conservative rates, trading spectrum efficiency for predictable performance. In this paper, we improve on this status-quo with NOSTRAdAMUS, a predictive LA framework which adds foresight to existing algorithms without replacing or redesigning them. NOSTRAdAMUS predicts whether a retransmission will occur in the next radio frame from recent HARQ history, and applies corrections to the Modulation and Coding Scheme (MCS) selected by the underlying policy. We benchmark several ML models and show that Gradient Boosting achieves 82.9% accuracy overall with high-confidence interventions that are correct 94.2% of the time, and an inference latency of 5.5 μs. We train the model based on data collected Over-the-Air (OTA) on the X5G testbed, using the open-source OpenAirInterface (OAI) 5G stack, NVIDIA Aerial, and COTS O-RAN Radio Units and User Equipments. The model is then deployed as a dApp, which we evaluate OTA as well as on various channels with hardware-in-the-loop channel emulators. This includes 3GPP TDL and CDL channels, SISO and MIMO configurations, and pedestrian and vehicular mobility. Our evaluation shows that without retraining, and across this variety of scenarios, the dApp augments two SOTA LA algorithms, and increases goodput by up to 71.5% while reducing retransmissions by up to 71.8%. This demonstrates the robustness and generalization capabilities of our approach.

Abstract

Wireless Communications

Explore similar work

Improving 5G AI-RAN MCS Selection by Predicting Retransmissions

RIDE: An Open Dataset and Benchmark for Train Delay Prediction

ML-Based Real-Time Downlink Performance Prediction in Standalone 5G NR Using Smartphones