The Design Of Synthetic Aperture Radiometer Simulation System On GPU Platform

The actual hardware of synthetic aperture microwave radiometer system is highly complex and it is hard to be accurately analysed, when using the computer simulation system to analysis the system error, model and algorithm establishment, and performance evaluation can highly reduce the developing risk develop period. There is a significant meaning in the research of synthetic aperture imaging simulation system.First, we introduce the development of the GPU through these years. Then describe the framework of the hardware and software of the CUDA which is the common computing of GPU. In order to extrude the feasibility of advantage using GPU in synthetic aperture microwave radiometer simulation system, this article divides the simulation system into four small modules according to the microwave radiometer detection process. Then analyzes the computational complexity and its changing process. From the analysis of the small modules, we can conclude that the whole system is highly complex. Simply compared the difference of random numbers occurring, FFT, matrix computing between GPU and CPU platform. Present the solution method of GPU speeding up computing. Thanks to the simulation system of microwave radiation detection process has a high possibility of parallelism, this article present the method of every small module on GPU platform. Finally do the experiments on GPU platform and only CPU platform and small Linux cluster platform. When Comparing the results, it certifies the GPU solves the computational complexity and data issues speedly and has a bright future.