CoRF enables efficient radio spatial mapping across new scenes
CoRF: Cross-Scene RF Synthesis by Learning Propagation and Preserving Array Physics
Networking and Internet Architecture
Summary
Mapping how radio signals spread in new locations usually requires lots of measurements and tuning for each spot. The authors created CoRF, a smart system that learns general rules about signal movement from many scenes and uses just a few measurements to predict spatial signal patterns in new, unseen places without extra training. CoRF separates the general signal behavior from the specifics of the antenna array and uses physics to accurately generate spatial radio data quickly. This approach was tested in many simulated environments and performed much better than previous methods.
What this means in practice
- •For wireless network engineers: Predict spatial radio signal patterns in new deployment sites quickly using limited target data without retraining models on each scene.
- •For autonomous vehicle system designers: Generate accurate spatial RF environment maps for unknown routes to improve vehicle localization and communication systems.
Authors
Kang Yang, Duaa Nakshbandi, Wan Du, Mani Srivastava
Abstract
Existing radio-frequency (RF) neural fields fit each scene separately, making new-scene deployment measurement- and optimization-intensive. This work studies amortized cross-scene spatial spectrum synthesis, where a shared model learns propagation across scenes and instantiates an unseen scene from sparse target-scene measurements without scene-specific training. To achieve this, CoRF separates learned scene-dependent propagation from the known receiver-array observation model. An unordered set of spectrum-only references conditions a canonical anchor field, producing arrival directions and query-dependent component powers for each query. Analytic array physics maps these components to the receiver covariance and then to the spatial spectrum. This factorization keeps the pretrained propagation model frozen, enabling synthesis at arbitrary query locations after a single reference-conditioning pass. A necessary local reference-capacity bound and an error decomposition further characterize the formulation. Across 35 simulated scenes spanning seven categories, CoRF outperforms the strongest baseline by 7.09 dB PSNR on unseen variants of represented scene categories and by 6.97 dB on entirely unseen scene categories.