Parallel Optimization of Spaceborne SAR Echo Simulation and Imaging Using OpenCL Based on GPGPU

A ground-based simulation system is necessary to verify the feasibility of real spaceborne Synthetic Aperture Radar (SAR) systems. Since echo simulation and image generation are computationally complex, parallel acceleration for SAR systems has been an active research area. However, most of the acceleration algorithms utilize compute unified device architecture (CUDA) as the programming platform, that works only on NVIDIA’s GPUs. This paper proposes a parallel optimization algorithm for spaceborne SAR echo simulation and imaging in Open Computing Language (OpenCL). In the echo simulation module, we optimize the Fast Fourier Transform (FFT), the calculation of range echoes, and the synchronization operation. In the imaging module, the data is decomposed to reduce the computational burden, and matrix transpose and phase factor multiplication are optimized using OpenCL. The architecture and steps needed to extract parallelism in the implementation of the algorithm for accelerated SAR echo simulation and imaging are described in detail. Benefiting from the better generality of OpenCL, this algorithm can be used to program General-Purpose Graphics Processing Units (GPGPU), i.e., CPUs, GPUs, and other types of processors. The performance is promoted by orders of magnitude compared with CPU-based implementation.