Wireless Physical Layer Foundation Models (WPFMs) offer a new approach to helping wireless networks understand and adapt to their radio environments. By learning from diverse radio-frequency data, these models can support a range of communication and sensing applications, enabling networks to better predict changing conditions, optimize performance, and automate network operations.

When combined with neuro-symbolic reasoning, which brings together AI learning capabilities and rule-based reasoning, WPFMs can also support more explainable and trustworthy AI-driven decisions. This white paper examines these emerging capabilities, their potential applications in future wireless networks, and the technical challenges and opportunities associated with their development and deployment.