Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Physics-informed line-of-sight learning for scalable deterministic channel modeling

Xiucheng Wang, Junxi Huang, Conghao Zhou, Xuemin Shen, Nan Cheng

Abstract

Deterministic channel modeling maps a physical environment to its site-specific electromagnetic response. Ray tracing produces complete multi-dimensional channel information but remains prohibitively expensive for area-wide deployment. We identify line-of-sight (LoS) region determination as the dominant bottleneck. To address this, we propose D$^2$LoS, a physics-informed neural network that reformulates dense pixel-level LoS prediction into sparse vertex-level visibility classification and projection point regression, avoiding the spectral bias at sharp boundaries. A geometric post-processing step enforces hard physical constraints, yielding exact piecewise-linear boundaries. Because LoS computation depends only on building geometry, cross-band channel information is obtained by updating material parameters without retraining. We also construct RayVerse-100, a ray-level dataset spanning 100 urban scenarios with per-ray complex gain, angle, delay, and geometric trajectory. Evaluated against rigorous ray tracing ground truth, D$^2$LoS achieves 3.28~dB mean absolute error in received power, 4.65$^\circ$ angular spread error, and 20.64~ns delay spread error, while accelerating visibility computation by over 25$\times$.

Physics-informed line-of-sight learning for scalable deterministic channel modeling

Abstract

Deterministic channel modeling maps a physical environment to its site-specific electromagnetic response. Ray tracing produces complete multi-dimensional channel information but remains prohibitively expensive for area-wide deployment. We identify line-of-sight (LoS) region determination as the dominant bottleneck. To address this, we propose DLoS, a physics-informed neural network that reformulates dense pixel-level LoS prediction into sparse vertex-level visibility classification and projection point regression, avoiding the spectral bias at sharp boundaries. A geometric post-processing step enforces hard physical constraints, yielding exact piecewise-linear boundaries. Because LoS computation depends only on building geometry, cross-band channel information is obtained by updating material parameters without retraining. We also construct RayVerse-100, a ray-level dataset spanning 100 urban scenarios with per-ray complex gain, angle, delay, and geometric trajectory. Evaluated against rigorous ray tracing ground truth, DLoS achieves 3.28~dB mean absolute error in received power, 4.65 angular spread error, and 20.64~ns delay spread error, while accelerating visibility computation by over 25.

Paper Structure

This paper contains 6 sections, 3 theorems, 17 equations, 24 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Results
  3. Discussion
  4. Methods
  5. Data availability
  6. Code availability

Key Result

Theorem 1

Given $m$ building vertices and $n$ evaluation points, D$^2$LoS reduces the per-transmitter LoS preprocessing complexity from $O(m \log m + n)$ to $O(M \log M + n)$, where $M$ is the number of predicted boundary vertices. In typical urban scenarios, $M \ll m$ and $M \log M \ll n$, so the effective p

Figures (24)

  • Figure 1: Qualitative comparison of RSS radio maps across four representative scenarios. Each row corresponds to one scenario from the RayVerse-100 test set. Columns from left to right: ground truth (GT), D$^2$LoS, No-Geom, RadioUNet levie2021radiounet, and RMTransformer li2025rmtransformer. D$^2$LoS reproduces the spatial power distribution with high fidelity. No-Geom captures the coarse pattern but exhibits blurred boundaries. RadioUNet and RMTransformer fail to reconstruct meaningful spatial structures.
  • Figure 2: Qualitative comparison of angular power spectra at selected receiver locations. Each row corresponds to one receiver position from a different test scenario. The horizontal axis is the azimuth arrival angle and the vertical axis is the received power in dBm. D$^2$LoS closely tracks the ground truth. No-Geom preserves dominant arrival directions but introduces errors in secondary peaks. RadioUNet and RMTransformer produce noisy angular profiles with incorrect peak positions.
  • Figure 3: Qualitative comparison of power delay profiles at selected receiver locations. Each row corresponds to one receiver position from a different test scenario. The horizontal axis is the propagation delay in nanoseconds and the vertical axis is the received power in dBm. D$^2$LoS accurately reproduces the delay-domain structure. No-Geom exhibits shifted peak positions. RadioUNet and RMTransformer produce delay profiles with spurious peaks at large delays.
  • Figure 4: Distribution of RSS prediction errors across 100 test scenarios. Violin plots show the per-scenario distribution of six metrics: a) MAE, b) RMSE, c) bias, d) MSE, e) NMSE, and f) Pearson correlation. Black diamonds indicate the mean. D$^2$LoS exhibits both the lowest mean error and the smallest variance. RadioUNet and RMTransformer show wide distributions with heavy tails.
  • Figure 5: Distribution of APS prediction errors across 100 test scenarios. Violin plots show: a) angular spread error, b) MDoA error, c) APS shape cosine, and d) APS shape RMSE. D$^2$LoS achieves concentrated low-error distributions. The No-Geom variant shows moderate degradation. RadioUNet and RMTransformer produce near-random angular predictions.
  • ...and 19 more figures

Theorems & Definitions (9)

  • Theorem 1: Computational complexity reduction
  • proof
  • Theorem 2: Pointwise diffracted field bound and shadow boundary continuity
  • proof
  • Remark 1: Forward-scatter smoothing
  • Theorem 3: Exactness of 2D$\times$2D LoS decomposition
  • proof
  • Remark 2: Non-convex footprints
  • Remark 3: Limitations of the 2.5D assumption