【作者】 岳利群；

【导师】 游雄；

【作者基本信息】 解放军信息工程大学 ， 地图制图学与地理信息工程， 2011， 博士

【摘要】 本文在学习,借鉴国内外相关研究成果的基础上,对分布式存储虚拟地理环境中的一些重要理论和关键技术以及相关算法做了较为深入的研究,通过大量的实验数据论证了论文提出的自适应空间数据模型和基于元数据的空间数据引擎,并在此基础上模拟构建了多级别多地区分布式存储虚拟地理环境,完成了原型平台的设计与开发,实现了地理空间数据的存储、管理、共享及可视化。主要研究内容和创新点如下:1.研究分析了现有分布式存储VGE的体系结构、数据资源、支撑技术和应用平台的特征及存在问题,建立了四层体系结构的分布式存储虚拟地理环境的架构,并从技术和应用两个层面,给出了基础层、资源层、服务层和应用层的内容构成和支撑技术,为分布式存储VGE的研究设计和应用服务提供了理论和方法指导。同时,探索并设计了基于网格、云模式和物联网模式下的虚拟地理环境的体系结构,为分布式存储VGE的发展提供了思路,也验证了本文提出的体系结构的扩展性和兼容性。2.建立了VGE元数据模型。结合分布式存储VGE的应用需求,建立了面向三维仿真模型、矢栅一体表达的元数据模型,实现了VGE元数据的动态管理、快速解析、缓存维护等8项技术,为分布式存储VGE元数据的规范化和标准化提供参考。3.建立了自适应的空间数据模型。提出了自适应空间数据模型建立的五大制约因素:数据、软件、用户、计算机和网络,通过改进基于球面Clipmap的数据模型和分布式存储VGE的数据模型,构建了自适应空间数据模型,解决了分布式存储VGE中空间数据自适应组织和管理的难题。其中改进后的球面Clipmap空间数据模型,空间数据利用率提高了2倍。4.构建了基于元数据的分布式空间数据引擎,解决了分布式存储空间数据的快速索引和提取问题。构建并分析了空间数据存储体系,提出了基于Linux构建多缓存服务器的数据服务模式,该服务模式不但提高了数据访问速度,还消除了Lustre文件系统不支持Windows客户端的瓶颈问题。在服务器快速定位、数据缓存设计、数据并行提取、Socket连接池等方面改进和实现了空间数据引擎的若干关键技术,最终构建了分布式存储VGE的空间数据引擎。5.实现了自适应空间数据可视化的多模式应用,建立了全球矢栅一体可视化框架和分布式空间数据可视化框架,探讨并实践了分布式空间数据可视化应用的加速技术,基于自适应空间数据模型生成了空间数据可视化多模式应用,验证了自适应空间数据模型的有效性。6.设计并实现了基于分布式存储的高效高可用性虚拟地理环境平台。对分布式存储虚拟地理环境平台(DSVGEP)中重点模块存储数据入库、元数据管理、自适应空间数据模型、空间数据服务引擎、可视化表达进行了说明,在此基础上对平台进行模拟部署与应用。经实验测试,单台Linux客户端能够同时支持100台可视化客户端的并行访问。更多还原

【Abstract】 In this paper ,on the basis of learning from domestic and foreign research results, the author do a more in-depth research on some important theories, key technologies and related algorithms of the distributed storage virtual geographic environment, In this article, adaptive spatial data models and metadata-based spatial data engine are demonstrated by a large number of experimental data. On this basis a multi-level and multi-area distributed storage virtual geographic environment is simulated and ultimately the prototype platform is completed practically. By this platform we can achieve the storage, management, sharing, and visualization of geography spatial data. This paper has conducted some research and practice in more detail, the main contents and innovations are as follows:1. Analyse the features and problems of the existing distributed storage VGE by summarizing the system architecture, data resource, supporting technologies and application platform. We construct four-level theoretical system architecture for distributed storage VGE, and give the content structure and supporting technologies for four layers, including base layer, resource layer, service layer and application layer. The four layers provide a theoretical and methodological guidance for the study design and application service of distributed storage VGE. We also explore and design the system architecture of VGE based on grid, cloud patterns, internet of things, which not only provide a good thought for the development of distributed storage VGE, but also demonstrate the expansibility and compatibility of the proposed architecture.2. Establish VGE metadata model. By combining with the application requirements of distributed storage VGE, we build the metadata model for three-dimensional simulation model and vector and raster data integrated expression, and realize eight key technologies for distributed storage VGE metadata, such as dynamic data management, rapid analysis and cache maintenance. All these work provide reference for normalization and standardization of distributed storage VGE metadata.3. Build the adaptive spatial data model. We propose five constrained factors to build the adaptive spatial data model, including data, software, user, computer and network. By improving the data model based on spherical Clipmap and distributed storage VGE data model, we construct the adaptive spatial data model. The data model help us solve the problems of adaptive spatial data organization and management in distributed storage VGE. By using Improved spatial data model based on spherical Clipmap, the spatial data utilization can increase by 2 times.4. We establish the metadata-based distributed spatial data engine to improve the indexing and retrieval speed of spatial data. We make construction and analysis of spatial data storage system, and propose a new data service model by the composition of several Linux cache servers. The service model not only accelerates the data access speed, but also eliminates the bottleneck of Lustre file system. We improve and achieve a number of key spatial data engine technologies, including server quick location, data cache design, data parallelism extraction, socket connection pooling, etc. Finally, we build a spatial data engine for distributed storage VGE.5. We achieve the multi-modal applications for adaptive spatial data visualization, and build a global vector and raster integrated visualization framework and distributed spatial data visualization framework. We also explore and practise some acceleration technologies of distributed spatial data visualization application. Finally, we generate multi-modal applications for spatial data visualization based on adaptive spatial data model, which verify the validity of the data model.6. We design and implement the system named DSVGEP (Distributed Storage Virtual Geographic Environment Platform). We make detailed description to the key modules of DSVGEP, including data storage, metadata management, adaptive spatial data model, spatial data service engine and visual expression. At last, we simulate a hardware deployment and complete the functional practice. By the experimental test, we find a single Linux client can support the concurrent access of 100 clients.更多还原