基于GPU并行射线跟踪的电波预测算法研究

Parallel Ray-tracing Algorithm of Radio-wave Propagation Prediction Based on GPU

针对利用射线跟踪算法进行小区电波预测的计算量非常大,运行时间较长,在实际工程中难以应用等的问题,本文在研究已有并行射线跟踪算法,即主从模式和对等模式并行射线跟踪的基础上,采用图形处理器(GPU)作为计算平台,结合GPU高并行处理能力,运用CUDA的编程框架,并在NVIDIA公司出厂的GEFORCE 9800GT上,实现了对城市小区的简单模型环境的电波预测。实验结果表明,基于GPU并行射线跟踪算法适用于城市小区的电波预测研究,该算法能大大节省通信时间,缩短计算时间,更方便更实用。

Ray-tracing algorithm for predicting radio-wave propagation in urban microcell needs a great deal of computation time. In order to more widely using this algorithm, we need to solve these problems. In this paper, based on previous research of parallel ray-tracing, namely Master/Slaves Model （MSM） and Peer to Peer Model （PPM）, present a new method which combines with the capability for parallel processing of GPU, and uses CUDA programming framework. In GEFORCE 9800GT produced by NVIDIA, we conduct some experiments to forecast radio-wave propagation in simple city environment. The results show that this method in the paper which could save communication-time and computation-time is more convenient than that uses several computes.