Radiomap Blind Prediction under Incomplete Observation: Error Characterization and Correctable Propagation-Prior Learning
Researchers introduce RadioDecomp to correct propagation-prior bias in radiomap prediction under incomplete observations.
The paper analyzes radiomap blind prediction when environment and base-station observations are incomplete, identifying the conditional mean as the optimal target and separating reducible approximation error from irreducible uncertainty. It argues that propagation priors both guide prediction and can bias the attainable predictor. RadioDecomp treats a prior-guided model as a correctable base and learns the remaining discrepancy, instantiated as feature-guided RadioFR and LoS-Shadow RadioLSR. Experiments across random, cross-configuration, and cross-environment settings show both improve on their bases.
- Incomplete observations leave radiomaps only partly determined by inputs.
- Error splits into reducible approximation and irreducible missing-physics uncertainty.
- RadioDecomp residual-refines a prior-guided base into RadioFR and RadioLSR.
- Both variants beat their bases across random, cross-configuration, and cross-environment tests.
From the original article
1 min read at arxiv.orgRadiomap blind prediction aims to infer radiomaps from observable representations of the propagation environment and base station configuration without field measurements. In practice, the observable representations are inherently incomplete. Thus, the target radiomap is not fully determined by the inputs when generalizing to unseen configurations or environments.
By Xiaojie Li · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.