MPI和Pthread混合编程在NCS算法中的设计和实现  

Parallel Imaging of MPI and Pthread Mixed Programming Based on Cluster

作　　者：田蕾[1] 范士明[1] 

机构地区：[1]北京卫星信息工程研究所,北京100086

出　　处：《中国空间科学技术》2010年第5期31-38,共8页Chinese Space Science and Technology

摘　　要：根据星载合成孔径雷达(Synthetic Aperture Radar,SAR)实时和准实时成像的需求,论文设计和实现了一种混合编程模式的非线性CS(Nonlinear Chirp Scaling,NCS)成像算法。此方法基于SMP(Symmetric Multiprocessing)集群系统,利用MPI(Message PassingInterface)和Pthread混合编程,实现了多进程和多线程的两级流水线并行成像处理。通过在HP刀片系统上的实验验证和结果分析,这种混合编程的流水线NCS并行成像算法既能较好适应前端系统传输一景原始回波数据时间不小于8s的情况,又能较好适应大规模原始回波数据堆积并且等待处理的情况。The mixed programming for Nonlinear Chirp Scaling（NCS） algorithm was designed according to the requirements of space borne Synthetic Aperture Radar（SAR） real-time and near-real-time imaging.The two-level pipelined parallel imaging of multi-process and multi-thread with Message Passing Interface（MPI） and Pthread was researched and implemented on SMP cluster.Through the experiments on HP cluster,it is proved that the mixed program imaging algorithm can not only adapt to the instance that the time of one block of original echo data transferred from the front system is more than or equal to eight seconds,but also the instance that a great number of original echo data has been already accumulated.

关 键 词：并行成像 混合编程 非线性CS算法 合成孔径雷达 卫星 

分 类 号：TP311.1[自动化与计算机技术—计算机软件与理论]