Research And Optimization On LBS Based POI Storage And Query

Along with the rapid development of mobile Internet,LBS based point of interest(POI)query prevails among all kinds of mainstream apps as a core function.Meanwhile,the dramatically increasing of LBS based POI data leads to the unsatisfactory demands which is divided in to two aspects: on the one hand,the space coordinates of the POI data processing ability with the existing relational database is relatively weak;on the other hand,relational database cannot effectively support lateral expansion and massive data processing.It is necessary to introduce a new technique,which can efficaciously support the spatial data storage,and is suitable for large-scale data processing.HBase,a distributed database based on Hadoop distributed framework,has inherent advantages in large-scale data processing and database horizontal extension.At the same time,Geohash supports the transformation from the two-dimensional coordinates into the one-dimensional strings,and remains the spatial characteristics of the original two-dimensional coordinates.Combined with Geohash,HBase is capable to effectively store and query the massive POI data.In view of the deficiency of the massive spatial data processing with the traditional relational database,combined with the characteristics of HBase's natural processing of massive data and characteristics of POI data,the researches are carried out from the index design of POI and region POI query respectively.Firstly,according to the HBase storage mechanism,the design principle of primary key and the feature of POI data,a new POI data index structure,namely GH-Index,is proposed combined with HBase and Geohash.Then,POI index parallel construction scheme of BulkLoad model is proposed based on GH-Index.Secondly,with the smallest rectangular approximation method,POI region search algorithm based on GH-Index is realized,which including rectangular region search algorithm and k nearest neighbor search algorithm.And the optimization of Region server-side filtering are proposed to speed up the query efficiency.Finally,the comparison test and analysis are carried out from the index construction of POI and the POI region query algorithm.Through a series of comparative experiments,the conclusions are drawn as follows: 1)the parallel constructing algorithm enhances the speed of index construction;2)region server-side filtering improves the efficiency of region query algorithm;3)comparison of index construction based on two dimensional latitude and longitude,region query algorithm based on GH-Index has better query efficiency and scalability.