aiqbench analysis
LatentWave introduces a Joint-Embedding Predictive Architecture (JEPA) to train wireless foundation models, aiming to overcome the low-level signal bias inherent in traditional masked reconstruction methods. By predicting masked regions in latent space rather than raw inputs, this approach yields more transferable representations for diverse downstream tasks like RF classification and 5G positioning without requiring task-specific fine-tuning. The study further demonstrates that specific masking geometries can induce inductive biases that favor either channel-related or signal-discriminative capabilities.
LatentWave: JEPA Pretraining for Wireless Foundation Models
arXiv:2606.06373v2 Announce Type: replace-cross Abstract: Wireless foundation models have emerged as a promising alternative to building separate models for each wireless task. However, existing approaches rely on masked input reconstruction, which can bias representations toward low-level signal details. In this paper, we propose LatentWave, a wireless foundation model pretrained using a Joint-Embedding Predictive Architecture (JEPA) on diverse wireless spectrograms and channel state information (CSI). By predicting masked regions in latent space, LatentWave learns representations that are more transferable out of the box across diverse downstream tasks. The proposed architecture employs per-channel patch embeddings with stochastic channel sampling during pretraining, allowing it to process variable antenna counts and improving usability across heterogeneous wireless configurations. We evaluate LatentWave on four downstream tasks: RF signal classification, 5G NR positioning, beam prediction, and LoS/NLoS classification, comparing against a masked-modeling baseline (WavesFM) pretrained on the same data. Additionally, we show that the masking geometry introduces a task-dependent inductive bias: frequency masking strongly favors channel-related tasks such as positioning and beam prediction, while region masking better preserves discriminability for signal classification.
Abstract:Wireless foundation models have emerged as a promising alternative to building separate models for each wireless task. However, existing approaches rely on masked input reconstruction, which can bias representations toward low-level signal details. In this paper, we propose LatentWave, a wireless foundation model pretrained using a Joint-Embedding Predictive Architecture (JEPA) on diverse wireless spectrograms and channel state information (CSI). By predicting masked regions in latent space, LatentWave learns representations that are more transferable out of the box across diverse downstream tasks. The proposed architecture employs per-channel patch embeddings with stochastic channel sampling during pretraining, allowing it to process variable antenna counts and improving usability across heterogeneous wireless configurations. We evaluate LatentWave on four downstream tasks: RF signal classification, 5G NR positioning, beam prediction, and LoS/NLoS classification, comparing against a masked-modeling baseline (WavesFM) pretrained on the same data. Additionally, we show that the masking geometry introduces a task-dependent inductive bias: frequency masking strongly favors channel-related tasks such as positioning and beam prediction, while region masking better preserves discriminability for signal classification.
Signal Processing (eess.SP); Artificial Intelligence (cs.AI)
arXiv:2606.06373 [eess.SP]
(or arXiv:2606.06373v2 [eess.SP] for this version)
https://doi.org/10.48550/arXiv.2606.06373 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 265 words
Related coverage
LatentWave: JEPA Pretraining for Wireless Foundation Models
arXiv · Jun 6, 2026
MultiPathFormer: Towards a Foundation Model for Multipath Wireless Propagation
arXiv · Aug 6, 2026
EEG-JEPA: Structured Latent Prediction for EEG Foundation Models
arXiv · Aug 5, 2026
V-JEPA Policy: Building Effective World-Action Models on Predictive Visual Latents
arXiv · Sep 30, 2026