PilotWiMAE pretrains an encoder on noisy pilots with factorized attention, 99% masking, patch-normalized reconstruction, scale loss, and AWGN curriculum to outperform supervised baselines in cross-frequency beam selection and channel tasks from 3.5 GHz pretraining to 28 GHz evaluation.
WiFo-2: a generalist foundation model unifies heterogeneous wireless system design
10 Pith papers cite this work. Polarity classification is still indexing.
10
Pith papers citing it
abstract
Emerging sixth-generation wireless systems are increasingly heterogeneous, with compatibility across diverse configurations, ubiquitous coverage, and expanded functionalities. Although deep learning has substantially benefited wireless system design, existing approaches are typically trained for specific system settings and scenarios with limited generalizability. Here we present WiFo-2, a space-time-frequency foundation model for unified wireless communications and sensing system design. Pretrained on a heterogeneous dataset of 11.6 billion channel state information (CSI) points, WiFo-2 learns generalized wireless representations across scenarios, configurations, and tasks, and exhibits scaling-law behavior. WiFo-2 achieves reliable and accurate zero-shot channel reconstruction, outperforming fully supervised task-specific models. With only 1% of the training samples required by supervised AI models, WiFo-2 achieves state-of-the-art performance across 9 distinct wireless tasks. A functional hardware prototype further demonstrates its real-world deployability and superior capability across diverse wireless tasks. This work provides a versatile wireless design framework and advances understanding of wireless channels.
years
2026 10verdicts
UNVERDICTED 10representative citing papers
CSI-CLIP++ uses CSI-CIR contrastive alignment to pretrain a channel encoder that improves beam prediction by up to 19.31 percentage points and supports positioning on DeepMIMO data across environments.
WiFo-MiSAC is a task-agnostic foundation model that unifies multimodal wireless signals via tokenization and self-supervised learning with SS-DMoE to achieve strong few-shot performance on beam prediction and channel estimation.
FARM combines a masked autoencoder with a diffusion-based decoder to construct aerial radio maps in both condition-based and condition-free settings, trained on a new large-scale 3D radio dataset (ARM-Omni).
AirFM-DDA reparameterizes wireless channel data into the delay-Doppler-angle domain and uses efficient window attention to achieve better zero-shot performance on channel prediction and estimation with lower compute cost.
A pre-trained interference-aware graph Transformer model for wireless resource allocation that achieves strong few-shot adaptation to new tasks and scenarios.
Adaptive 3D-RoPE adapts rotary positional encoding to wireless channel physics via learnable 3D frequencies and dynamic CSI control, yielding up to 10.7 dB NMSE gains in scale extrapolation and 1 dB in zero-shot tasks.
A unified framework for CSI-native foundation models incorporates scale-aware exposure, physical coordinates, and correlation-bounded attention, reporting over 4 dB NMSE gains in zero-shot tasks and 36.6% spectral efficiency improvement with 7% pilot overhead.
RAFC adaptively composes multi-level hidden features from wireless foundation models via a task-driven router, outperforming standard adaptation methods on four tasks with under 50K extra parameters.
Surveys adaptation of foundation models to wireless tasks across off-the-shelf, wireless-native, and agentic paradigms for 6G PHY intelligence and network autonomy.
citing papers explorer
-
PilotWiMAE: Pilot-Native Representation Learning for Wireless Channels
PilotWiMAE pretrains an encoder on noisy pilots with factorized attention, 99% masking, patch-normalized reconstruction, scale loss, and AWGN curriculum to outperform supervised baselines in cross-frequency beam selection and channel tasks from 3.5 GHz pretraining to 28 GHz evaluation.
-
CSI-CLIP++: A Scalable Channel Foundation Model for Wireless Communication via CIR-CSI Consistency
CSI-CLIP++ uses CSI-CIR contrastive alignment to pretrain a channel encoder that improves beam prediction by up to 19.31 percentage points and supports positioning on DeepMIMO data across environments.
-
WiFo-MiSAC: A Wireless Foundation Model for Multimodal Sensing and Communication Integration via Synesthesia of Machines (SoM)
WiFo-MiSAC is a task-agnostic foundation model that unifies multimodal wireless signals via tokenization and self-supervised learning with SS-DMoE to achieve strong few-shot performance on beam prediction and channel estimation.
-
FARM: Foundational Aerial Radio Map for Intelligent Low-Altitude Networking
FARM combines a masked autoencoder with a diffusion-based decoder to construct aerial radio maps in both condition-based and condition-free settings, trained on a new large-scale 3D radio dataset (ARM-Omni).
-
AirFM-DDA: Air-Interface Foundation Model in the Delay-Doppler-Angle Domain for AI-Native 6G
AirFM-DDA reparameterizes wireless channel data into the delay-Doppler-angle domain and uses efficient window attention to achieve better zero-shot performance on channel prediction and estimation with lower compute cost.
-
A Graph Foundation Model for Wireless Resource Allocation
A pre-trained interference-aware graph Transformer model for wireless resource allocation that achieves strong few-shot adaptation to new tasks and scenarios.
-
Adaptive 3D-RoPE: Physics-Aligned Rotary Positional Encoding for Wireless Foundation Models
Adaptive 3D-RoPE adapts rotary positional encoding to wireless channel physics via learnable 3D frequencies and dynamic CSI control, yielding up to 10.7 dB NMSE gains in scale extrapolation and 1 dB in zero-shot tasks.
-
Towards CSI-Native Foundation Models: A Channel-Adaptive Roadmap for 6G
A unified framework for CSI-native foundation models incorporates scale-aware exposure, physical coordinates, and correlation-bounded attention, reporting over 4 dB NMSE gains in zero-shot tasks and 36.6% spectral efficiency improvement with 7% pilot overhead.
-
A Unified Adaptive Feature Composition Framework for Multi-Task Generalization in Wireless Foundation Models
RAFC adaptively composes multi-level hidden features from wireless foundation models via a task-driven router, outperforming standard adaptation methods on four tasks with under 50K extra parameters.
-
Foundation Models for Wireless Communications: From PHY Intelligence to Network Autonomy
Surveys adaptation of foundation models to wireless tasks across off-the-shelf, wireless-native, and agentic paradigms for 6G PHY intelligence and network autonomy.