Research On Vector Space-time Query Algorithm Based On Redis

With the rapid development of 3S(RS,GIS,GPS)technology,spatio-temporal data has become one of the important engines for the development of various industries.The era of "Internet + spatio-temporal big data" has come to an end.Multi-scale,multi-dimension,and high-precision vector spatio-temporal data are exponentially growing.How to efficiently store,index,and apply the spatio-temporal data has become a hot issue to be solved.Vector data is an important form of geospatial data,and its efficient organization and management is one of the key issues in the application of spatial data.The traditional relational database considers such characteristics as transaction,paradigm constraints and so on,and can no longer meet the large-scale,high-concurrency spatio-temporal data application requirements.NoSQL database has been widely used by academia and business community for its good performance in the storage and management of big data complex information because of its distributed,extensible and non-ACID features.Based on this,this article uses the Redis database to deeply study the storage model of vector spatio-temporal data,spatio-temporal index,spatio-temporal query and so on.The research content has the following three aspects:(1)A Redis-based vector data storage model was constructed.In order to satisfy the efficient storage and query of vector spatiotemporal big data,using Redis several common types of key-value data,a hierarchical indexing mechanism for vector spatiotemporal data is proposed.Firstly,it discusses the vector-oriented big data storage model at home and abroad.Second,it designs a vector space-time hierarchical storage model,and on this basis organizes the vector data feature objects in each layer.As a theoretical basis and research basis for the follow-up work of this paper.(2)A Redis-based vector spatio-temporal query algorithm is proposed.In view of the lack of time dimension considerations in the application of space-time data vector query,a two-level index model of vector time-space based on Redis is constructed.First,the space-time data space is divided,and Hilbert encodes the spatial feature objects.Second,the modules with the same Hilbert value are integrated to form multiple spatial data sets.Each module is a set of layers referenced at different times.Finally,the use of Redis is appropriate.The key-value data type organizes the model.The experiment is expanded on a real data set.Compared with the traditional spatial database Oracle Spatial,the spatial-temporal range query and the parallel empty space-time query test are performed.The experimental results show that the proposed algorithm reduces the query time by 4.5 times,and the concurrency performance is better.It is verified that Redis is more suitable for efficient query and concurrent processing of massive spatio-temporal data.(3)A Hilbert-R tree hierarchical indexing algorithm combining time and clustering results is proposed.Tree index and clustering technology have become the hot topics in the study of spatial indexing technology.However,when constructing tree index,it is easy to cause node overlap and dead space problems.A hierarchical index model based on Hilbert-R tree is proposed.First,based on Redis' s rich key-value data type,the time index is built on the periodic dataset.Secondly,the Hilbert partition space is further processed by clustering.Finally,the Hilbert-R tree hierarchy is constructed using the clustering results and time attributes.Index structure.A spatio-temporal query processing scheme based on Hilbert-R tree was designed to perform data-intensive and space-time query tests.After verification,the proposed algorithm is not affected by the data distribution,which can effectively reduce the time overhead,shorten the query time by an average of 25%,and have a relatively small amount of access to tree nodes.