Survey Maps Wireless Foundation Models for 6G Networks.

A new comprehensive survey provides an in-depth review of wireless foundation models (WFMs) and their transformative potential for AI-native sixth-generation (6G) wireless networks. Unlike traditional deep learning models, WFMs are designed to learn generalized representations from vast, heterogeneous wireless data, enabling scalable and transferable intelligence across diverse communication tasks. The survey establishes a taxonomy for WFMs, categorizing them by architecture, pre-training paradigms, and applications. It details representative architectures, self-supervised pre-training strategies, parameter-efficient adaptation methods, relevant datasets, benchmarks, and evaluation methodologies. Furthermore, the paper explores emerging applications in physical-layer signal processing, network intelligence, and cross-layer optimization, while also addressing key challenges such as data availability, generalization, interpretability, efficient edge deployment, and standardization, outlining future research directions for trustworthy 6G intelligence.