Window Function Optimization-based Query Of Trajectory Data

With the development of wireless communication and mobile computing, mobile devices gradually plays a more and more important role within various applications in our daily life. Nowadays, the mobile devices around us often have positioning and wire-less communication ability. Therefore, location-based services and uploading the data to server have been possible. Track and record the trajectory data is particularly important. Trajectory data contains the movement of the moving object. With the rapid increase of trajectory data, the demand for trajectory analysis is also growing, includes the military, transportation, science, logistics and social service, etc. The purpose of this paper is to reduce the positioning error and sampling error that exist in the trajectory data and imple-ment the trajectory data analysis and mining.Main contributions of this paper are as follows:·Trajectory Data Preprocessing In view of positioning error in the trajectory data, we remove the abnormal points by using the mean filter and match the trajectory data onto the city road network by using the hidden markov map-matching model, in order to improve the accuracy of trajectory data.·Window Function Optimization Modeling the consumption model of window function in the process of sequence calling in the PostgreSQL, we find out the bot-tleneck. We propose the optimization methods for the aggregation functions and MAX/MIN functions under the window function. Than, we prove the validity of the two optimization methods through theory and experiment.·The construction of query for trajectory data We generate the road network mode of the trajectory data through the preprocessing for trajectory data and con-struct the road network analysis and query system by using the window function.In summary, we preprocess the trajectory data based on the trajectory filter and map matching technology to reduce the effects of positioning error, storage trajectory data by using relational database and reduce the effects of sampling error in the data analysis and mining of trajectory data by using the window function. Under the background of the sharply increase of the trajectory data, the implementation framework of window func-tions in relational database has the bottleneck. In this paper, we propose the optimization methods for the aggregation functions and MAX/MIN functions under the window func-tion and build the query system of road network.