A Hybrid Interpolation Framework for Sparse RF Digital Twin Measurements

Abstract

Digital Twins (DTs) are expected to be adopted in beyond-5G and 6G wireless systems to support realistic performance evaluation and system validation. However, highfidelity DT calibration typically requires either dense measurement campaigns or computationally intensive deterministic simulations, which limit scalability and practical deployment. This paper proposes a hybrid simulation-assisted interpolation framework for DT channel calibration that combines sparse real measurements with low-cost ray-tracing simulations. Measured and simulated channel responses are first statistically aligned to compensate for systematic discrepancies, after which missing channel samples are reconstructed from a reduced measurement set by interpolating the measurement-simulation residual. Two interpolation strategies are investigated: Inverse Distance Weighting, serving as a low-complexity distance-based baseline, and Ordinary Kriging, which provides a statistically grounded reconstruction through data-adaptive weighting. The proposed framework is validated using two complementary measurement configurations: an indoor scenario based on a mechanically synthesized large virtual antenna array, and an outdoor scenario employing a rotating horn antenna, extending the interpolation problem to the angular domain. Reconstruction performance is assessed under varying measurement compression levels using RMSE, NMSE, and Pearson correlation. The results demonstrate that accurate channel reconstruction can be achieved with substantial measurement reduction (over 80 percent), enabling scalable and cost-efficient calibration of radio-frequency DT scenarios.