Research On The Analysis Method Of The Characteristics Of Human Mobility Trajectory Based On MapReduce

In recent years,with the popularity of GPS-enabled mobile device and the development of positioning technology,all these applications will generate a lot of trajectory data.For example,navigation system will record location information when it is turned on.The step software of smartphone will record the locus that the user walked.The map software can locate the user's current locations and provide the navigation function.However,it is heavy burden of the massive trajectory to store,transmit and process even if it is good for human mobility feature analysis.In the face of the above-mentioned problems,this thesis has carried out the research on human mobile trajectory feature analysis method based on MapReduce,including parallel trajectory compression method,human mobile trajectory feature analysis algorithm and the analysis software prototype system based on it.Aiming at the problem of data redundancy caused by the massive human trajectory data set,In this thesis,a parallel trajectory compression method based on MapReduce is proposed.At the same time,because parallelism will cause the correlation of trace points to be destroyed before and after segmentation point.The method is as follows:Firstly,the trajectory is divided with two segmentation methods in which the segmentation points are interleaving.Then,the trajectory segments are assigned to different nodes for parallel compression.Lastly,the compression results are matched and merged.The performance test and analysis results show that the proposed method can not only increase the compression efficiency significantly,but also eliminate the error which is caused by the destructive problem of correlation.Then,for the real human trajectory data set after compression,in order to analyze the trajectory dataset efficiently,this thesis presents the analysis algorithm based on MapReduce to analyze contact time,encounter interval time and flight distance.Firstly,the trajectory dataset will be segmented,and then each segmentationwill be divided into different nodes for feature analysis.The result of performance test shows that the method proposed in this thesis can improve the efficiency of trajectory feature analysis.Finally,the software of human trajectory feature analysis based on MapReduce is realized in this thesis.The system is composed of Hadoop cluster and application platform,and the proposed methods based on MapReduce are applied in this thesis,such as a parallel trajectory compression method,a parallel contact time analysis method,a parallel encounter interval time analysis method and a parallel flight distance analysis method.The experiments show that the system can significantly improve the processing efficiency when dealing with a large number of trajectory data.Finally,the analysis results are displayed on the application platform.The research of this thesis is of great significance to deal with massive real human movement trajectory data,which greatly improves the efficiency of data storage,transmission and analysis.