Research On The Density Query Technology For Moving Objects

In recent years,with the widely application of mobile technology,researches on density query technology for moving objects are receiving more and more attention.Density queries are defined as querying the dense regions that include more than a certain number of moving objects,which are widely used in crowded area discovery,transportation control,animal behavior monitoring and resource planning.Although there are great achievements and progress about queries in spatial and temporal space,technologies on density queries for moving objects need more deep researches.And in the Big Data era,the problem of sparse data,which leads to uneven data distribution,brings new challenges to density queries.Based on the problems above,this thesis aims at the researches on density queries for moving objects and on sparseness of multidimensional data.The mainly works are below.Firstly,in terms of density query region selection,if the query range is the whole spatial space,the number of moving object is large,and the query space is irregular,this will lead to high complexity of query running time.In this thesis,we apply the method of frequent area mining based on grid and the research on the whole space focuses on frequent grid regions,as these frequent regions are more likely to become the dense areas.And we build Markov Model on these frequent areas to calculate the probability of moving objects at certain positions in the near future.Next,in the realization of spatial area density queries,this thesis presents a solution to the conventional density query.Then for the problem of low accuracy of query method,a density query algorithm based on group relation is proposed by studying the similarity between the moving object trajectory,the space object is grouped,and the problem of the density query of a single object is converted to a study of each object group.We run the algorithms on real GPS trajectory datasets and test the accuracy of two density query algorithms and the impact of the number of moving objects on density query time.Finally,we apply the method to solve the sparseness of multidimensional data.The density status and type of some areas can be obtained through GPS and monitoring videos.However the data is sparse and can reflect the distribution of a few areas at some time.Characteristics of every region can be picked up through POI data,check-in data and road network data,and then be used to fill the density region data with sparseness.We evaluate the fill correctness with root-mean-square error and mean absolute error,and the experiment result show the good fill effect so as to obtain the density situation of all regions.In conclusion,this thesis builds a complete query.framework based on a meaningful query problem,solving the problem of data quality.We use the sparse data and different data sources to fill the missing data efficiently to obtain a more precise prediction result.