Acceleration of shooting and bouncing ray method based on OptiX and normal vectors correction

Area of Science:

• Computational electromagnetics
• High-performance computing

Background:

• The shooting and bouncing ray (SBR) method is crucial for predicting electromagnetic scattering from complex geometries.
• Traditional SBR methods face challenges with computational efficiency and accuracy, especially for curved surfaces represented by facets.

Purpose of the Study:

• To develop a novel, efficient SBR method leveraging modern GPU hardware.
• To enhance the accuracy of ray tracing for faceted surfaces.
• To accelerate the computation of radar cross-section (RCS) for arbitrary geometries.

Main Methods:

• Implementation of a new SBR algorithm utilizing the OptiX API and hardware-accelerated ray tracing cores (RT Cores) on NVIDIA RTX graphics cards.
• Incorporation of a normal vector correction technique to mitigate ray tracing errors on faceted surfaces.
• Parallelization of far-field integration calculations across multiple GPU cores to enhance computational speed.

Main Results:

• The corrected ray paths closely match theoretical predictions, validating the normal vector correction method.
• The algorithm accurately predicts the RCS of arbitrary faceted geometries.
• The proposed method achieves a speedup of 60 times compared to traditional kd-tree-based SBR methods.

Conclusions:

• The developed OptiX-based SBR method with normal vector correction offers a significant advancement in computational efficiency and accuracy for electromagnetic scattering analysis.
• This approach effectively utilizes GPU computing resources for rapid RCS prediction.
• The method provides a robust solution for analyzing complex, faceted targets in electromagnetics.