BVH-Accelerated Ray Tracing for High-Frequency Electromagnetic Backscattering
Marco Pasquale, Andong Hu, Luca Pennati, Ivy Peng, Stefano Markidis · 2026
Read BVH-Accelerated Ray Tracing for High-Frequency Electromagnetic Backscattering by Marco Pasquale, Andong Hu, Luca Pennati, Ivy Peng, Stefano Markidis (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-09243.txt.
Introduction
Pasquale and colleagues present a shooting and bouncing rays method for high-frequency backscattering from metal objects. Full-wave solvers grow infeasible for electrically large problems as frequency rises. It couples ray transport with a physical optics integral over ray tubes, using a bounding volume hierarchy to cut the search. Accuracy is controlled by a sampling rule that limits phase aliasing. It runs on NVIDIA and AMD GPUs with MPI, validates against Mie solutions for a sphere, and predicts radar cross-section for an aircraft.
Catalog record
| CREATOR | Marco Pasquale, Andong Hu, Luca Pennati, Ivy Peng, Stefano Markidis |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45607 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.09243
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
BVH-Accelerated Ray Tracing for High-Frequency Electromagnetic Backscattering Marco Pasquale, Andong Hu, Luca Pennati, Ivy Peng, Stefano Markidis Abstract As computational complexity in electromagnetics increases with frequency, full-wave solvers become computationally infeasible for electrically large problems. To address this limitation, we present a shooting and bouncing rays (SBR) method for efficiently modeling electromagnetic backscattering of metallic objects in the high-frequency regime. The method couples multi-reflection geometrical-optics ray transport with a physical optics surface integral discretized over ray tubes. To reduce the massive ray-surface intersection search space, we use a bounding volume hierarchy (BVH) and organize the computation as a trace–integrate pipeline. The ray tracing generates hit data, and the physical optics integral is evaluated over valid intersections only. Numerical accuracy is controlled through an incident-ray sampling rule that mitigates phase aliasing in the discretized physical optics integration. The method is accelerated on NVIDIA and AMD GPUs and parallelized with MPI. We validate against analytical Mie solutions for a perfectly electrically conducting (PEC) sphere and demonstrate applicability to a complex aircraft geometry for monostatic radar cross-section prediction. Keywords: Shooting and bouncing rays ⋅ \cdot Bounding volume hierarchy ⋅ \cdot Electromagnetic scattering ⋅ \cdot Radar cross section ⋅ \cdot GPU acceleration Note. This manuscript is an extended version of the paper accepted at the 26th International Conference on Computational Science (ICCS 2026). Software. The source code is available at: https://github.com/marco-pas/SagittaSBR 1 Introduction Computational electromagnetics (CEM) includes both numerical and asymptotic methods for predicting electromagnetic fields governed by Maxwell’s equations in realistic geometries and materials. It is widely used across applications such as antenna and microwave component design, wireless propagation and channel modeling, radar and remote sensing, and scattering analysis for complex platforms [ 21 , 22 , 25 , 18 , 6 , 5 , 17 ] . Within this broad spectrum of applications, different modeling regimes call for different tools. Full-wave methods are preferred when resonance, diffraction, or strong coupling dominate, while asymptotic or hybrid approaches can be more efficient when the geometry is large compared to the wavelength and field interactions become strongly directional. When the characteristic size of a structure D is much larger than the wavelength \lambda (i.e., D/\lambda\gg 1 ), full-wave discretizations must resolve many wavelengths across the geometry, leading to very fine meshes, large linear systems, and often impractical memory and runtime requirements. This challenge is becoming increasingly important in emerging 6G scenarios extending toward sub-terahertz frequencies [ 11 ] . In the high-frequency limit \lambda\ll D , however, many scattering effects are dominated by specular interactions and slowly varying surface currents, motivating asymptotic approaches that replace a global field solve with ray-like transport and coherent accumulation of scattering contributions. Ray tracing provides a computational framework for this regime. In computer graphics, ray tracing models transport by launching rays, computing intersections with scene geometry, and applying local interaction rules at each hit. In HF electromagnetics, a closely related structure arises. Rays approximate geometrical-optics (GO) propagation and reflection, while electromagnetic boundary conditions determine the reflected field and phase accumulation along a path. The scattered field is then obtained by coherently assembling contributions over the illuminated surface, commonly through a physical-optics (PO) surface-current integral [ 14 ] . Shooting and bouncing rays (SBR) combines these ideas by shooting a discretized set of incident rays toward the target and bouncing them via specular reflections to capture multiple interactions. This is followed by a PO accumulation step. SBR was introduced in the electromagnetics literature to enable practical radar cross section (RCS) prediction for geometries that were difficult to treat with full-wave tools, such as open-ended cavities [ 12 ] , and has since been developed and applied broadly as a ray-based alternative in the HF regime [ 19 ] . While foundational concepts date back to classical works, recent advancements in GPU architectures and parallel hierarchies have drastically revitalized ray-based methods for contemporary CEM applications. SBR requires repeated ray–surface intersection queries across large ray ensembles and multiple reflections. In addition, the number of ray–triangle tests grows rapidly with ray count and bounce depth. As a result, naïve all-against-all ray–triangle testing becomes infeasible at the sampling densities needed for stable PO accumulation. This work targets that dominant bottleneck by accelerating ray–triangle intersection with a bounding volume hierarchy (BVH), a hierarchical spatial data structure that reduces intersection cost by culling large portions of the mesh using bounding boxes [ 13 ] . When edge and shadowing effects are important, SBR is often augmented with high-frequency diffraction theories such as GTD, and refinements such as UTD or PTD that improve behavior near shadow boundaries and enforce continuity between illuminated and shadowed regions [ 7 , 9 , 1 , 8 , 24 , 3 ] . From a numerical perspective, the PO surface integral is evaluated through a discretization induced by the incident ray grid, so insufficient sampling can introduce artifacts in the accumulated field. However, in high-fidelity multiple-reflection simulations on complex triangulated targets, the dominant practical challenge is typically computational. The goal of this work is to develop and assess a computationally efficient SBR method for HF electromagnetic backscattering in the presence of complex geometry and multiple reflections. Specifically, we (i) formulate a ray-tube discretization of the PO surface integral driven by a controlled incident-ray sampling rule, (ii) employ a BVH to reduce the cost of repeated ray–triangle queries during multi-bounce transport, and (iii) organize the procedure as a trace–integrate pipeline in which ray traversal produces compact intersection data that is subsequently consumed by PO field evaluation. The method is accelerated on NVIDIA and AMD GPUs and parallelized with MPI by distributing angular sweep parameters across ranks, enabling large parameter studies for electrically large targets. We validate accuracy against analytical Mie solutions for a perfectly conducting sphere and demonstrate applicability to a complex aircraft configuration for monostatic radar cross-section prediction. 2 Related work Ray-based high-frequency electromagnetics. Computational treatments of PO and related integral evaluations are widely documented in the CEM literature [ 14 ] . Shooting and bouncing rays (SBR) is a representative GO+PO technique that combines multi-reflection ray transport with a PO accumulation step to predict quantities such as radar cross section. The method was introduced in the electromagnetics literature to enable RCS analysis of geometries that were challenging for full-wave methods at the time, including open-ended cavities [ 12 ] , and later developments positioned ray-based approaches as practical tools for HF scattering and related applications [ 19 ] . Beyond RCS, ray-based solvers and hybridizations appear in problems such as electromagnetic coupling and bidirectional coupling analysis [ 21 , 22 ] , propagation and channel modeling [ 25 , 18 , 6 ] , antenna placement and design-space exploration [ 5 ] , and mixed full-wave/ray-tracing strategies that use ray methods where appropriate while retaining full-wave fidelity in subregions [ 17 ] . A known modeling limitation of classical GO+PO SBR is the absence of diffraction mechanisms, which can be important near edges, tips, or shadow boundaries. When needed, diffraction effects are incorporated via HF corrections such as the geometrical theory of diffraction (GTD) [ 7 ] and refinements such as the uniform theory of diffraction (UTD) [ 9 , 1 , 8 ] or the physical theory of diffraction