Point2Radio: A Foundation Model for Cross-Scene Radio Fields from Material-Aware Point Clouds

2026-07-30

AI Summary

This work addresses the limited generalization of existing wireless signal field modeling methods, which struggle to reuse propagation structures across diverse scenarios. The authors propose the first point cloud–based foundation model for wireless fields, leveraging material-aware point cloud encoding and transmitter-conditioned embeddings to construct a unified scene representation that supports multi-task query decoding. Without relying on voxelization or explicit ray tracing, the model enables real-time prediction of three-dimensional path gain and power angular spectrum on a single GPU. Evaluated on a large-scale dataset comprising 337 scenes, it achieves a mean absolute error of only 0.871 dB in path gain prediction—representing a 76.7% improvement over a UNet baseline—and demonstrates strong capabilities for efficient fine-tuning and multi-task extension.

Abstract

High-fidelity radio fields are typically simulated for every scene--transmitter configuration or fitted separately to each scene, failing to exploit propagation structures shared across environments. We present Point2Radio, a foundation model that learns a transferable propagation prior from multiple environments. Given a material-aware point cloud and a transmitter (TX) setting, a common encoder produces a TX-conditioned scene representation that can be queried at arbitrary receiver (RX) locations. Task-specific query decoders map this representation to different radio quantities, e.g., three-dimensional (3D) path-gain (PG) fields and power angular spectra (PAS). At inference for a new scene, the model uses only a material-aware point cloud and transceiver queries, running in milliseconds on a single GPU without meshes or explicit path tracing. We evaluate PG prediction on a scene-disjoint split of a 337-scene corpus containing 86,272 TX-conditioned fields. Point2Radio achieves 0.871 dB mean absolute error (MAE), reducing error by 76.7% relative to a same-split UNet-style baseline. The same encoder also supports PAS prediction via a task-specific decoder. Experiments further show that light target-scene fine-tuning improves adaptation to a specific environment.

Problem

• radio fields
• cross-scene generalization
• propagation modeling
• material-aware point clouds
• foundation model

Innovation

• foundation model
• material-aware point cloud
• radio field prediction
• cross-scene generalization
• path-gain field