GPU-based Shooting And Bouncing Ray Method Of Target Electromagnetic Scattering

Shooting and Bouncing Ray(SBR) method is a high frequency approximation method that use ray tubes to simulate electromagnetic waves propagation and scattering situation in the target surface, which is very suitable for the complex target electromagnetic computing problems. But tracking and electromagnetic calculation of each tube are very time consuming because of the large ray tube numbers, greatly limits the application of SBR method in electromagnetic scattering calculation. This paper mainly discusses how to make full use of the existing technology to improve the computational efficiency of SBR method, implements a method based on Bounding Volume Hierarchy(BVH) binary tree acceleration data structures and GPU-based CUDA parallel computing techniques that can significantly improve the computational efficiency of SBR method. The main work of this paper is as follows:1. The implementation of traditional SBR method is introduced. First generate ray tube, split out the initialization ray tube on the virtual hole. Second track and update of each ray tube tracking, ray and target triangle surface intersection and tube electromagnetic field intensity update. Then use the Physical Optics method to calculate the RCS of each ray tube which is updated, target RCS is equal to the sum of all ray tube RCS. Traditional SBR method is used to calculate several models, and the calculation results of SBR method compared with the calculation results of FEKO confirmed the effectiveness of the method and can be applied to multiple scattering of electromagnetic computing problem.2. Then introduces the BVH accelerate tree to improve the computing efficiency of SBR method. First introduces the performance of four types of bounding box, and Aligned Axis Bounding Box(AABB) is used to surround the target surface element. Then introduces middle division and Surface Area Heuristic(SAH) these two kinds of space division methods, and use those space division methods to divide the target surface dimension repeatedly until the BVH tree is constructed. At last describes in detail the stack traversal method which is used in the process of ray tracking. The numerical calculation results show that: BVH tree can significantly reduce the tracing time of each ray tube, acceleration effect of BVH tree is proportional to the depth of the constructed tree. What’s more, the calculation precision of the SBR method is not depends on the size of each target triangle element, but depends on the size of each ray tube.3. Next introduces GPU-based CUDA parallel computing techniques to further improve the computational efficiency of SBR method. In order to deal with BVH-SBR method on the GPU, first introduces the CUDA programming model. Then describes in detail how the huge amount of ray tube are processed on the GPU, use the rope to enhance BVH tree and introduces the process of the rope adding for each node. At last discuss in detail the non-stack tree traversal method which is used in the process of ray tracking, can reduce the unnecessary internal node traversal. The numerical calculation results show that: GPU can improve the calculation efficiency of a large amount of ray tube tracking, and within GPU computing capability, the more the number of ray tube, the more acceleration effect provided by the GPU.4. Finally GPU-based SBR method is used to the target of radar stealth optimization analysis, mainly analyze a closed octahedral mast model. First introduces four kinds of radar stealth performance evaluation parameters. Then calculate the RCS of original mast at one scattering and multiple scattering situations, and analysis the influence of multiple scattering for mast RCS. At last analyze the influence of four different parameters mast models and its own radar stealth performance.