Research On Optimization Of Trajectory Data Query In Big Data Environment

The trajectory data refers to the data with spatiotemporal characteristics formed by sampling the process of the moving object.With the continuous development of location acquisition technology and the large-scale deployment of acquisition equipment,more and more moving object trajectory data has been generated.These can be used for transportation planning,urban planning,interest recommendation,etc.,and have a wide significance and important value.If use trajectory data for analysis,fast and efficient query is essential,and the efficiency of the query is usually related to the data storage method and index structure.The storage method and index structure are closely related to the quality of the data itself.The quality of the data is related to the original data and data preprocessing.For data analysis,the original data can no longer be optimized,but careful data preprocessing can make subsequent analysis work faster and more effective.Therefore,this paper has carried out detailed research work on trajectory segmentation,trajectory query and trajectory index structure,and designed and implemented the segmentation algorithm,index structure and query algorithm.There are three specific aspects as follows:1.Considering the time-space characteristic of fixed-point trajectory data,a dynamic threshold travel recognition method for fixed point trajectory data is proposed.At first,use hierarchical clustering to determine the spatial-temporal multiple granularity parameters which relate to the threshold.Then count historical records according to parameters to calculate the threshold corresponding to each parameter.Last,do trajectory segmentation with spatialtemporal threshold to get the precise travel recognition result.Experiment based on fixed point trajectory data of real world city shows that using spatialtemporal dynamic threshold method to do travel recognition to fixed point trajectory data is superior to the traditional stable and single threshold method on accuracy and coverage.2.The spatiotemporal index structure for three types of typical trajectory data query is given.First,according to the traditional trajectory query types,two types of trajectory queries,point query and range query,are summarized.Secondly,according to the trajectory segmentation results,a new type of trajectory query is proposed;due to the difference in query performance under different indexing methods,this article sets different index structures separately,R-tree index,Z3 index,and Z3+ index,for the point query,range query,and segment query.So that when performing trajectory query,a suitable index structure can be used to improve the efficiency of query.3.Designed and implemented a prototype system for trajectory data query and management in a big data environment.In this paper,Hadoop is selected as the basic environment for trajectory data storage and query.Hbase is selected as the database for storing original trajectories and standard trajectories.It is developed based on Geomesa's open source framework and implements three types of typical queries and corresponding index construction.At the same time,based on the system's experiments on the relevant real trajectory data sets,it is verified that the work in this paper has better performance in trajectory segmentation and trajectory query than the traditional work,and it also shows the effectiveness and applicability of this work.