Uncertainty Quantification of Radio Wave Propagation over Irregular Terrains Using Adaptive Polynomial Chaos Expansion

TL;DR

This paper introduces an improved adaptive polynomial chaos expansion method coupled with a parabolic wave equation to efficiently quantify uncertainties in radio wave propagation over irregular terrains, achieving accurate predictions with fewer simulations.

Contribution

The paper presents a novel adaptive polynomial chaos expansion approach that enhances uncertainty quantification for irregular-terrain radio wave propagation, outperforming existing methods in accuracy and efficiency.

Findings

Accurately predicts mean and percentile range of path loss with only 30 simulations.

Outperforms standard and sparse PCE methods in accuracy and robustness.

Maintains low errors and variability as sample size increases.

Abstract

Accurate modeling of radio wave propagation over irregular terrains is crucial for designing reliable wireless communication systems in such environments, yet uncertainties in the antenna configuration are not quantified within deterministic models. In this paper, we present, to the best of our knowledge, the first uncertainty quantification (UQ) study of realistic antenna configurations for irregular-terrain propagation. An adaptive polynomial chaos expansion (APCE) method is improved and coupled with a two-way parabolic wave equation (PWE) method to address this problem efficiently. The polynomial basis is extended according to variance contributions and terminated by a composite criterion combining validation error and sample-to-basis ratio, enabling stable coefficient estimations via least-square regression without additional regularization. Convergence analysis shows a monotonic error decay with increasing training samples, producing compact, low-interaction models and improved accuracy and robustness over the previous APCE methods. For two realistic terrain profiles, the proposed method accurately predicts the mean and the 5th-95th percentile range of the path loss, matching Monte Carlo (MC) references using only 30 PWE simulations. Using a fixed sampling budget, APCE outperforms standard and sparse PCE, with the largest gains observed for the 5th and 95th percentile estimates; as the sample size increases, APCE maintains low errors with reduced trial-to-trial variability.