基于多处理板联合SAR成像的分布式计算模拟系统研究

合成孔径雷达（SAR）卫星的探测范围有限，全轨利用率不高，并且单个卫星的计算和存储资源有限，难以完成大量的星载合成孔径雷达数据处理。本文拟采用多处理板联合的数据处理方法，针对星上分布式计算构建一种模拟系统，以提高星上数据处理效率，验证多处理板联合的数据处理方法在高效利用卫星计算与存储资源、提升单个卫星全轨利用率等方面的优点。本系统基于距离多普勒（range Doppler，RD）成像算法的多卫星分布式SAR实时处理方法，并以现场可编程门阵列（FPGA）芯片为核心构建了分布式计算模拟系统。不同于传统的单个卫星RD算法处理过程，该系统将处理过程分为三个阶段。每个阶段内运算任务被合理地分配给不同的数据处理单元。利用高分三号卫星（GF-3） SAR原始数据进行成像处理，以检验方法和系统的性能。

Distributed computing simulation system based on a multi-processing board combined with SAR imaging

Synthetic-aperture radar (SAR) satellites have a limited detection range, low full-orbit utilization rates, and limited computing and storage resources on a single satellite, making performing many spaceborne SAR data-processing tasks difficult. This paper uses the data-processing method of multi-processing boards to build a simulation system for onboard distributed computing to improve the efficiency of data processing. It verifies the advantages of multi-processing boards'' data-processing methods in the efficient use of satellite computing and storage resources and in improving the full-orbit utilization rate of a single satellite. This system is based on the range-doppler (RD) algorithm of the multi-satellite distributed SAR real-time processing method and on the field-programmable gate array (FPGA) chip. Unlike the traditional single-satellite RD algorithm, this system divides the process into three stages. The computing tasks are appropriately assigned to different data-processing units within each stage. Original data of the Gaofen-3 SAR satellite are used for imaging processing in order to test the performance of the method and system.