Abstract:This paper presents the Large Wireless Model (LWM) -- the world's first foundation model for wireless channels. Designed as a task-agnostic model, LWM generates universal, rich, contextualized channel embeddings (features) that potentially enhance performance across a wide range of downstream tasks in wireless communication and sensing systems. Towards this objective, LWM, which has a transformer-based architecture, was pre-trained in a self-supervised manner on large-scale wireless channel datasets. Our results show consistent improvements in classification and regression tasks when using the LWM embeddings compared to raw channel representations, especially in scenarios with high-complexity machine learning tasks and limited training datasets. This LWM's ability to learn from large-scale wireless data opens a promising direction for intelligent systems that can efficiently adapt to diverse tasks with limited data, paving the way for addressing key challenges in wireless communication and sensing systems.

What problem does this paper attempt to address?

This paper attempts to address several key challenges in wireless communication and sensing systems, especially those related to high - dimensional signal processing, complex optimization problems, large - scale wireless overhead requirements, and complex network management. Specifically: 1. **Limited labeled data**: Traditional deep - learning methods usually require a large amount of labeled data sets, but in wireless networks, such data are often scarce and difficult to collect. 2. **Complex spatio - temporal dependencies**: Existing deep - learning models (such as CNN and RNN) have difficulty effectively capturing time - dependent and long - range dependencies when handling wireless communication and sensing tasks. 3. **Insufficient generalization ability across multiple wireless environments**: Traditional modeling techniques (such as statistical models and optimization - based methods) usually rely on simplified models or features of specific scenarios and cannot generalize well to diverse dynamic environments. To solve these problems, the authors propose the Large - scale Wireless Model (LWM), which is a fundamental model specifically designed for wireless channels. The main features of LWM are as follows: - **Task - agnostic general - purpose feature extractor**: LWM generates general, rich, and context - aware channel embeddings (features) through pre - training, which can enhance the performance of various downstream tasks. - **Self - supervised pre - training**: LWM is pre - trained in a self - supervised manner on a large - scale wireless channel data set, so it can learn complex structural relationships without a large amount of labeled data. - **Transformer architecture**: LWM utilizes the multi - head attention mechanism in the Transformer architecture and can capture complex spatial and temporal relationships in wireless channel data. - **Patch processing**: The wireless channel is divided into patches so that LWM can efficiently capture local and global patterns while maintaining computational efficiency. - **Masked Channel Modeling (MCM)**: By masking part of the channel data and requiring the model to reconstruct this data, LWM can learn more robust feature representations during the pre - training stage. Overall, LWM aims to overcome the key challenges in wireless communication and sensing systems by providing a powerful fundamental model, especially in cases where labeled data is limited, spatio - temporal dependencies are complex, and generalization across multiple environments is required. This provides a new direction for the development of future intelligent systems, enabling them to adapt more effectively to various tasks and perform well with limited data.

Large Wireless Model (LWM): A Foundation Model for Wireless Channels

Large Multi-Modal Models (LMMs) as Universal Foundation Models for AI-Native Wireless Systems

WirelessLLM: Empowering Large Language Models Towards Wireless Intelligence

Towards a Wireless Physical-Layer Foundation Model: Challenges and Strategies

The Power of Large Language Models for Wireless Communication System Development: A Case Study on FPGA Platforms

Large Language Model (LLM)-enabled In-context Learning for Wireless Network Optimization: A Case Study of Power Control

Large Language Models in Wireless Application Design: In-Context Learning-enhanced Automatic Network Intrusion Detection

Large Language Models (LLMs) Assisted Wireless Network Deployment in Urban Settings

The Role of Federated Learning in a Wireless World with Foundation Models

Leveraging Large Language Models for Wireless Symbol Detection via In-Context Learning

WirelessAgent: Large Language Model Agents for Intelligent Wireless Networks

WDMoE: Wireless Distributed Large Language Models with Mixture of Experts

A Human Body Model for Efficient Numerical Characterization of UWB Signal Propagation in Wireless Body Area Networks

Large Language Models for Networking: Workflow, Advances and Challenges

Token Embedding Frequency Embedding Time Embedding Antenna Embedding Position Embedding

Large AI Model Empowered Multimodal Semantic Communications

Personalized Wireless Federated Learning for Large Language Models

WDMoE: Wireless Distributed Mixture of Experts for Large Language Models

DeepChannel: Wireless Channel Quality Prediction Using Deep Learning

Beam Prediction based on Large Language Models

Large Language Model Enabled Semantic Communication Systems