集群环境下的VLBI软件相关处理方法研究
文献类型：学位论文
| 作者 | 张冬 |
| 学位类别 | 硕士 |
| 答辩日期 | 2007-06-15 |
| 授予单位 | 中国科学院上海天文台 |
| 授予地点 | 上海天文台 |
| 导师 | 郑为民 |
| 关键词 | 甚长基线干涉测量 集群 软件相关处理 并行算法 消息传递 |
| 中文摘要 | As a radio astronomical observation technology with very high spacial resolution, Very Long Baseline Interferometry (VLBI) has been widely used in the satellite tracking and orbit determination activities. VLBI correlator is the kernel device of pre-processing of VLBI. There are two kinds of correlator, hardware correlator and software correlator. At present, the software correlator based on high performance computers becomes a research emphasis in VLBI technology. The software correlator has already made a success of the deep space exploration, and it begins to be used in astrophysics and space geodesy data processing. Increasing VLBI stations and bandwidth demands the augment of data rate. How to build a software correlator with high speed、real-time and parallel processing ability becomes a highlight of VLBI technology field. . At present, the cluster is the most popular architecture of the High Performance Computing (HPC) fields. It can solve the complicated scientific computing and intensive data processing problems. Using HPC cluster to solve researching and engineering problems is a mainstream. In first three chapters, the backgrounds of VLBI technology, and the software correlators, as well as the principle of cluster computing were presented. Then the cluster platform of the software correlator was discussed. This thesis focused on the processing method and realization of a VLBI software correlator under a small cluster system. Firstly, the principle of current software correlator was presented. Secondly, a cluster system based on the current computing resource was introduced. A correlation algorithm with two level parallel model (MPI+Pthread) was carried out in the VLBI FX software correlator. Finally, the parallel program’s performance and the optimization methods were discussed. According to the research result, a VLBI software correlator running on a cluster system with 3 nodes（8 CPUs in each node）was developed with the ability of 4 stations, 6 baselines data correlation. Under the condition of 4 stations, the correlator was able to achieve 118Mbps/station data processing rate. |
| 英文摘要 | 甚长基线干涉测量技术（VLBI）具备高精度测角能力，已被应用于对人造卫星及深空航天器的跟踪观测及定位。VLBI相关处理机是VLBI技术中数据预处理的核心设备，它分为硬件相关处理机和基于通用计算机平台的软件相关处理机。近年来，基于高性能PC或服务器平台的软件相关处理机技术已经成为VLBI技术领域的研究热点。软件相关处理技术已经在深空探测领域获得了极大的成功，并且开始在天体物理和空间大地测量方面得到应用。 目前，观测站数目的增多与观测带宽的增大，带来了数据速率的提高。软件相关处理技术如何向实时、高速、大规模并行处理是VLBI领域中一个重要的具有前沿性的研究方向。 在高性能计算领域，集群已作为主流体系结构而被广泛采用。它具备突出的特点，如能够应对复杂的科学计算，海量数据的并行处理等。通过构建高性能集群来解决复杂的科研及工程问题，已成为当今一个主流的解决方案。 本文前三章阐述了VLBI技术背景、国内外软件相关处理系统现状，集群计算原理，以及集群系统在软件相关处理系统中的国内外应用现状。 本文重点研究在小规模集群环境下VLBI软件相关处理方法与实现技术。首先分析目前软件处理机的基本原理、特点。其次利用现有资源，组建多节点集群计算系统，提出MPI和Pthread相结合的两级并行模式的相关处理算法，并在此基础上编写并行相关处理软件。最后对并行相关处理软件在集群中的处理性能进行分析，并提出进一步的优化方法。 基于本文研究内容，研究并开发了3节点、每节点8个CPU的集群计算系统，以及具备4台站、6条基线、并行相关处理能力的VLBI软件相关处理软件。此软件在并行处理4台站、每台站128Mbps速率的数据中，能够达到118Mbps/台站的相关处理速率。 |
| 语种 | 中文 |
| 公开日期 | 2011-07-01 |
| 页码 | 69 |
| 源URL | [http://119.78.226.72//handle/331011/14688]  |
| 专题 | 上海天文台_中国科学院上海天文台学位论文 |
| 推荐引用方式 GB/T 7714 | 张冬. 集群环境下的VLBI软件相关处理方法研究[D]. 上海天文台. 中国科学院上海天文台. 2007. |
入库方式： OAI收割
来源：上海天文台
浏览0
下载0
收藏0
其他版本
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。