Enabling AI-Native Mobility in 6G: A Real-World Dataset for Handover, Beam Management, and Timing Advance

A commercially deployed network dataset capturing pedestrian, bike, car, bus, and train mobility for AI/ML-driven handover, beam management, and timing advance prediction

Mannam Veera Narayana; Rohit Singh; Deepa M. R; Radha Krishna Ganti· 2026· DOI 10.48550/arXiv.2605.12453

The core problem

The high interruption time and measurement report overhead under user equipment (UE) mobility, especially in high-speed 5G use cases, remain critical challenges. AI/ML techniques for beam management and mobility procedures have been proposed to address these issues, but they rely heavily on simulated data that do not accurately reflect real deployment behavior or user traffic patterns. There is an utmost need for realistic datasets under various conditions. This work presents a dataset collected from a commercially deployed network across various modes of mobility (pedestrian, bike, car, bus, and train) and at multiple speeds to depict real-time UE mobility. The dataset focuses primarily on handover (HO) scenarios, aiming to reduce HO interruption time and maintain continuous throughput during and immediately after HO execution. It includes timing advance (TA) measurements at various signaling events such as RACH trigger, MAC CE, and PDCCH grant, which are typically missing in existing works. The paper provides a detailed description of dataset creation, experimental setup, data acquisition, and extraction, along with an exploratory analysis focused on mobility, beam management, an

The dataset encompasses a wide range of mobility scenarios, from pedestrian to train, at multiple speeds, providing a comprehensive view of UE mobility in a commercial 5G network. Key measurements include handover events, beam management parameters, and timing advance values at critical signaling points. The exploratory analysis reveals patterns in handover interruption time and throughput during and after handover execution. The dataset includes TA measurements at RACH trigger, MAC CE, and PDCCH grant, which are often missing in existing datasets. These results enable the evaluation of AI/ML models for predicting TA and optimizing handover procedures. The dataset's realism, derived from a commercial deployment, ensures that models trained on it can generalize better to real-world conditions. The inclusion of multiple mobility modes and speeds allows for robust testing across diverse scenarios.

The high interruption time and measurement report overhead under user equipment (UE) mobility, especially in high-speed 5G use cases, remain critical challenges. AI/ML techniques for beam management and mobility procedures have been proposed to address these issues, but they rely heavily on simulated data that do not accurately reflect real deployment behavior or user traffic patterns. There is an utmost need for realistic datasets under various conditions. This work presents a dataset collected from a commercially deployed network across various modes of mobility (pedestrian, bike, car, bus, and train) and at multiple speeds to depict real-time UE mobility. The dataset focuses primarily on handover (HO) scenarios, aiming to reduce HO interruption time and maintain continuous throughput during and immediately after HO execution. It includes timing advance (TA) measurements at various signaling events such as RACH trigger, MAC CE, and PDCCH grant, which are typically missing in existing works. The paper provides a detailed description of dataset creation, experimental setup, data acquisition, and extraction, along with an exploratory analysis focused on mobility, beam management, and TA. Multiple use cases are discussed, including training and evaluating AI/ML models for TA prediction.

The dataset was collected from a commercially deployed network to capture real-world UE mobility. Data acquisition involved multiple mobility modes: pedestrian, bike, car, bus, and train, with varying speeds. The experimental setup focused on handover scenarios, recording key signaling events and measurements. Timing advance (TA) measurements were captured at RACH trigger, MAC CE, and PDCCH grant events. The data extraction process ensured inclusion of parameters relevant to handover interruption time, throughput continuity, and beam management. The methodology emphasizes realistic deployment behavior and user traffic patterns, contrasting with simulated datasets. The dataset creation process is described in detail, covering experimental setup, data acquisition, and extraction procedures. The resulting dataset supports AI/ML model development for mobility procedures, beam management, and TA prediction.

The dataset addresses the gap in realistic data for AI-native mobility in 6G. By providing real-world measurements across various mobility modes and speeds, it facilitates the development of AI/ML techniques that reduce handover interruption time and measurement report overhead. The inclusion of timing advance measurements at multiple signaling events enables accurate TA prediction, which is crucial for maintaining uplink synchronization. The dataset supports multiple use cases, such as training and evaluating AI/ML models for TA prediction, beam management, and handover optimization. The exploratory analysis highlights the importance of real deployment data in understanding inference of AI/ML models. Future work can leverage this dataset to design more efficient mobility procedures for high-speed scenarios. The dataset's focus on handover scenarios and throughput continuity aligns with the goals of 6G to provide seamless connectivity. The availability of such data accelerates the adoption of AI-native mobility solutions.

CS practitioners and researchers

Opening member content…