LWM-TF: A Wireless Foundation Model on 5G NR Time-Frequency Resource Grids
Abstract
Received OFDM grids preserve both pilot and payload observations, yet wireless foundation models are rarely pre-trained on them. We introduce LWM-TF, a 9.58M-parameter factorized masked autoencoder pre-trained on complex 5G NR uplink grids taken after the FFT and before channel estimation. Each block attends first across OFDM symbols and then across resource blocks and receive antennas, and the mask follows the same two axes. Because the payload phase is unpredictable, the model reconstructs the normalized log-power of each masked token and predicts its scale separately; ablations show that the two terms are complementary and that this target outperforms complex-signal reconstruction. The encoder is pre-trained on 850,496 physical uplink shared channel (PUSCH) slots from a 10^6-slot dataset simulated with the NR transmit chain under various channel conditions. With the encoder frozen, linear probes classify modulation, predict CRC success, and detect UE mobility. On a separately generated test set, 16 labeled slots per class give 0.85 modulation accuracy, 0.82 CRC area under the ROC curve (AUROC), and 0.73 mobility AUROC, ahead of a supervised CNN on every paired draw. With 10,000 labels the probes reach 0.94, 0.96, and 0.92, well above engineered receiver statistics, and detect mobility where the CNN stays at chance. Without any impairment labels, distances in the frozen feature space flag unseen waveform impairments with 0.78 macro AUROC, against 0.68 for receiver statistics. A single pre-trained encoder thus provides receiver features that transfer across tasks, most strongly when labels are scarce.
Related
- Adversarial Water-Filling: Theory, Algorithms, and a Domain-Specific Wireless Foundation Model
- Quantum Annealing for Resource Allocation in Next-Generation Wireless Communication Networks: A QUBO Modeling Guide
- A fixed-time stable limit theorem for non-normalized functionals on irregular time grids
- An efficient and rigorous criterion for the admissible time step of deforming grids in generalized Lagrangian and ALE methods
- Polynomial preserving recoveries of edge element method on Cartesian grids for the time-harmonic Maxwell equations with large wave number
BibTeX
@misc{kim2026lwmtf,
title = {{LWM-TF: A Wireless Foundation Model on 5G NR Time-Frequency Resource Grids}},
author = {Namhyun Kim and Sadjad Alikhani and Ahmed Alkhateeb},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.21013},
eprint = {2610.21013},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.21013},
note = {CC BY-NC-SA 4.0},
}
RIS
TY - UNPB
TI - LWM-TF: A Wireless Foundation Model on 5G NR Time-Frequency Resource Grids
AU - Namhyun Kim
AU - Sadjad Alikhani
AU - Ahmed Alkhateeb
PY - 2026
DA - 2026/10/14
PB - arXiv
DO - 10.48550/arxiv.2610.21013
UR - https://arxiv.org/abs/2610.21013
AB - Received OFDM grids preserve both pilot and payload observations, yet wireless foundation models are rarely pre-trained on them. We introduce LWM-TF, a 9.58M-parameter factorized masked autoencoder pre-trained on complex 5G NR uplink grids taken after the FFT and before channel estimation. Each block attends first across OFDM symbols and then across resource blocks and receive antennas, and the mask follows the same two axes. Because the payload phase is unpredictable, the model reconstructs the normalized log-power of each masked token and predicts its scale separately; ablations show that the two terms are complementary and that this target outperforms complex-signal reconstruction. The encoder is pre-trained on 850,496 physical uplink shared channel (PUSCH) slots from a 10^6-slot dataset simulated with the NR transmit chain under various channel conditions. With the encoder frozen, linear probes classify modulation, predict CRC success, and detect UE mobility. On a separately generated test set, 16 labeled slots per class give 0.85 modulation accuracy, 0.82 CRC area under the ROC curve (AUROC), and 0.73 mobility AUROC, ahead of a supervised CNN on every paired draw. With 10,000 labels the probes reach 0.94, 0.96, and 0.92, well above engineered receiver statistics, and detect mobility where the CNN stays at chance. Without any impairment labels, distances in the frozen feature space flag unseen waveform impairments with 0.78 macro AUROC, against 0.68 for receiver statistics. A single pre-trained encoder thus provides receiver features that transfer across tasks, most strongly when labels are scarce.
ER -