Research On Trajectory Query System

How to store,compute,query and analyze trajectory data is becoming a hot topic since it is propagating quickly.This thesis studies the collection,transmission,conversion,storage and query of massive trajectory data.Furthermore,explore faster query solution for the larger volume and larger space-time scale in order to improve the timeliness of data mining in trajectory applications.The study intends to provide underlying service support for smart cities.The main contents of this article are listed as follows:(1)Utilize the popular big data technologies such as Flume,Kafka,Flink,and HBase to build a trajectory query system consisting of data collection,transmission,conversion processing,storage and retrieval.A system with high fault tolerance,high performance,scalability is implemented preliminarily to meet the trajectory data processing requirements.(2)Design and implement Geohash encoding,Google S2 encoding,and ST-hash encoding algorithms in Flink.By different scenarios,different index schemes are designed for the latitude and longitude information and time information of trajectory data,which provides a basis for improving the efficiency of trajectory query.(3)Design a trajectory data storage model based on HBase database in the application background of moving object close contact query.A quadtree-based time range-oriented moving object trajectory query and an octree-based space-time range-oriented moving object trajectory query row key are designed in order to achieve faster trajectory data query in larger volume.(4)The feasibility of the trajectory query system is verified.The prototype experiment of acquisition,transmission,conversion processing and storage was carried out using 1000 trajectory data records.The performance test of moving object trajectory query oriented to time and spatiotemporal range is carried out by 1 billion trajectory data.In the space-time range query,the query times of Geohash,Google S2,and ST-hash are compared.The experimental results show that the trajectory query system designed can perform the acquisition,transmission,calculation and storage of trajectory data.Time range oriented moving object trajectory query can complete in 500 milliseconds.