An Efficient SAR Processor Based on GPU via CUDA
Dept. of Electron. Eng., Shanghai Jiao Tong University, Shanghai, China
2009 2nd International Congress on Image and Signal Processing (2009) Publisher: IEEE, Pages: 1-5
@conference{liu2009efficient,
title={An Efficient SAR Processor Based on GPU via CUDA},
author={Liu, B. and Wang, K. and Liu, X. and Yu, W.},
booktitle={Image and Signal Processing, 2009. CISP’09. 2nd International Congress on},
pages={1–5},
year={2009},
organization={IEEE}
}
A novel and efficient Synthetic Aperture Radar (SAR) processor is introduced in this paper. This new processor is implemented on the Graphics Processing Unit (GPU). GPU is traditionally used for graphics rendering, but in recent years, it has rapidly evolved as a highly-parallel processor with tremendous computation capability and ultra-high memory bandwidth. The algorithm of the new SAR processor is developed via Compute Unified Device Architecture (CUDA) which is a popular GPU programming environment. The imaging results of the simulated data show that the resolution and the peak sidelobe ratio of this processor agree with the theoretical values. The time of processing on the actual data shows that this processor is more than 18 times as fast as quad-core CPU-based processor using OpenMP. The high efficiency of this GPU-based processor provides a promising way to solve the complicated problems of high resolution SAR systems.
March 10, 2011 by hgpu
