| The development of scientific research and application of geospatial sience is often accompanied by the development of computer information technology and made great progress. At the same time, it also challenges the computer information technology. With the rapid development of global internet, sensor technology and earth observation technology, the geospatial science has entered a new stage. Spatial information service is changing from digital earth to the smart earth era. The challenges include four aspects:data intensity, computing intensity, concurrent access intensity and spatiotemporal intensity. Tranditional geographic information public service platform has been unable to meet the requirements of smart city. On the other hand, the information technology has entered the era of cloud computing and big dta. New computing models and techniques provide new solutions to those problems encountered in the development of geospatial science. Spatial cloud computing is a cloud computing paradigm that is driven by geospatial sciences. Spatiotemporal cloud computing platform is the concrete realization of spatial cloud computing. It is one key part of construction of smart city. This paper is launched around spatialtemporal cloud computing. Leveraging spatial cloud computing to support the four intensities of geospatial sciences, such as big spatial data storage and management, high performance spatial computing and how to construct a spatiotemporal cloud computing platform. The main research contents and results are as follows:(1) The framework of spatiotemporal cloud computing platforma) The concepts and framework of cloud computing:Describe the relevant concepts of cloud computing in detail, including the service model and classification of cloud computing. The four cloud service model is public cloud, private cloud, Community cloud and hybrid cloud. Then discuss the popular cloud computing layers:IaaS, PaaS and SaaS.b) Spatial cloud computing (SCC or SC2):discuss the concept and framework of spatial cloud computing. Compare the difference between SCC and tranditional cloud computing, such as DaaS to deal with big spatial. Then study how spatial cloud computing can help resolving down the 4 intensities problems of spatial science.c) Propose a framework of spatiotemporal cloud computing for smart city. Describe the 4 layers detail, include IaaS, DaaS, PaaS and Saas of the framework. And a virtual service (SCCVS) was designed to optimize the resource scheduling and auto scaling.(2) several key technologies to construct spatiotemporal cloud computing platformThere are many key technologies of building traditional cloud computing such as SOA, virtualization technology, multi tenant etc. To spatial cloud computing, there are 3 new key technologies for geospatial science:resource scheduling based on spatiotemporal regularities, load banlance and auto scaling, service chaining.a) Resource scheduling based on spatiotemporal regularities:review the virtualization technologies and resource scheduling for cloud computing. Then put forward the resource scheduling framework and algorithm on spatiotemporal cloud computing. Build a Quad scheduling model by immune clone algorithm to achieve the optimal for multi spatiotemporal goals.b) Load balance and auto scaling:The load balance and auto scaling of cloud computing are discussed in detail. The main technologis such as simple threshold rule, Q-learning and queuing theory are studied. Then we choose the simple threshold rule for auto scaling on spatiotemporal cloud computing platform and propose a strategy based on spatiotemporal princples and implement the algorithm.c) Service chaining on SCC:reviewed the concepts and combination technology of service chaining follow OGC geospatial service standards. Then introduce the work flow technology for service chain. Propose two model of service chain for spatiotemporai cloud computing:integrated dynamic computing resource service chain and optimization based on spatiotemporal principles.(3) Big spatial data management based on spatiotemporai cloud computing platformThe concept and key technology of big data are discussed in detail. Propose a DaaS for big spatial data and present an application which is build on DaaS and for storing and distributing of geoconditions census image data.a) Big data and key technologies:The concept of big data and challenge to IT are discussed. Key technologies including distributed file system, efficient processing and analysis are described in detail. Then review NoSQL technology and the architecture and characteristics of HBASE.b) Aspects of big spatial data. Compared with the traditional internet big data, the big spatial data has a more obvious mulit-source heterogeneous, spatiotemporal attributes, multi-dimension and distribution characteristics.c) A DaaS of spatiotemporal cloud computing:propose a DaaS architechure which uses the hybrid storage structure of HDFS+HBASE+RDBMS to provide storage service capability for big spatial data.d) DaaS provide Storage of spatial data:discusse the concept model of spatial data storage. Propose the storage mechanism of grid data and vector data based on DaaS.e) Predefined multiple indices mechanism (PMIM), propose a spatiotemporai index optimization mechanism of spatial data storage and access. Design a PMIM for access spatial data by spatiotemporal index.(4) High performance computing(HPC) for spatial analysis based on spatiotemporal cloud computing platforma) Review the high performance computing. Even though the performance enhancements increasing of super computer, the computing paradigm evolves to a new stage in cloud computing.b) High performance computing based on SCC:discuss HPC of spatiotemporal cloud computing, include virtualization platform, cloud management software, storage management, etc. Propose an architecture for high performance computing based on spatiotemporal cloud computing platform and Hadoop cluster, which makes the hadoop cluster more scalabler.c) A high performance spatiotemporal statistics analysis system (Hadoop-Geostatistics GIS):propose a system based on spatiotemporal cloud platform and Hadoop to support efficient spatiotemporal statistical analysis. Which is used to statistics variety of spatial statistical indexes. The calculation flow and MapReduce algorithm are implemented. The experiment results show that the system has significant performance improvement, and can be horizontal extension.(5) A prototypes system(SCCP) based on spatial cloud computingBased on the result of this paper, a prototypes system spatiotemporal cloud computing platform (SCCP) is proposed, including preliminary design of the IaaS, DaaS, PaaS and SaaS. The experiment has been carried on to verify some archievements of the research. |
PDF Full Text Request