Research On The Acceleration Method Of Ray Tracing Algorithm In The Application Of Complex Electromagnetic Environment

As a high frequency approximation method of computational electromagnetics,the ray tracing method can effectively solve the problems of simulation and calculation in the complex electromagnetic environment.However,the traditional ray tracing method has the problems of low efficiency and long time consuming.As a result,it is necessary to speed up the algorithm.Generally speaking,there are two ways to accelerate the ray tracing method.One is improving the speed of the algorithm itself(hardware acceleration)and the other is reducing the useless intersection calculation(algorithm acceleration).This article starts from these two ideas respectively.In terms of the hardware acceleration,we used the lateral-ray tube model,which is with higher accuracy,in the ray tracing method.Besides,we adopted the CUDA language to implement the GPU hardware acceleration,including the time consuming intersection calculation,the reflection calculation and the diffraction calculation.What is more,we analyzed the way of improving the efficiency of the data transferring between the CPU and the GPU.The numerical experiment demonstrates that the GPU-based implementation can significantly improve the computational efficiency of lateral-ray tube model about 17 times faster,while providing the same accuracy as the CPU-based one.In terms of the algorithm acceleration,we used the KD-tree accelerating structure.Moreover,we proposed a fast pruning algorithm based on the information of the current ray and the node.The experiment shows that the improved method is 17.3%faster than the traditional KD-tree accelerating structure,and 107.8%faster than the grid accelerating structure.