Research On Spatio-temporal Index Model And Retrieval Methods Based On HBase

Spatio-temporal data is a kind of muti-dimensional data with space attribute,time attribute and other attributes.With the development and extensive application of wireless communication,positioning and sensors,spatio-temporal data ushers in explosive growth in the type of data,the growth rate and data size.Because of the limited storage capacity and scalability,traditional relational databases can hardly meet the storage and management requirements of massive spatio-temporal data.As a representative of non-relational databases,HBase has better scalability which provides a new research direction for the storage and management of spatio-temporal data.Currently,the spatio-temporal indexing and retrieval technology based on HBase still exists many problems.On the one hand,spatio-temporal index based on tree structure is complex and with poor concurrency;and on the other hand,the index with specific structure can support very limited query process and with poor scalability.In this thesis,by analyzing the characteristics of HBase and spatio-temporal data and based on the theory of Hilbert curve space filling curve,secondary index HBase and MapReduce parallel computing technology framework,we researched the HBase–based spatio-temporal index model and retrieval method.The detailed description of this thesis is as follows:(1)Designed spatio-temporal storage model based on HBase: Well-formed storage structure is beneficial to improve the system's throughput,and also provides convenience for the design of index.This thesis expounded the design philosophy of spatio-temporal storage model in Rowkey and Column Family respectively,and gave brief analysis of its performance.(2)Proposed spatio-temporal index model based on HBase: Using the feature that the Rowkey of HBase are sorted in lexicographic order,combined with the Hilbert curve space filling technology and time coding algorithm,we designed spatio-temporal index on Rowkey.Considering that the application requirements are always changing,and the needs of spatio-temporal data muti-attribution digging,we designed the Region level secondary index on the Extend column family.Our spatio-temporal index is more appropriate for the needs of cloud computing;and on the other hand,our secondary index has better performance in terms of data consistency and scalability.(3)Designed spatio-temporal query algorithms based on HBase: The index is designed to serve a query,and the query algorithm must be designed to adapt to the specific index structure.This thesis designed the spatio-temporal query parallelization framework.Combined with the framework and spatio-temporal index,we proposed three spatio-temporal query algorithms.(4)Designed and developed a prototype system in order to evaluate the temporal and spatial overheads of spatio-temporal index creation and the performance of spatio-temporal query algorithms.The experiment results show that the proposed spatio-temporal index model and retrieval method based on HBase are effective.