Research On Travel Destination Prediction And Privacy Protection Based On Trajectory Data

Research on urban computing about prediction or estimation using all kinds of data can solve urban problems,such as more and more traffic accidents,traffic congestion,environmental degradation,noise pollution,energy consumption,and so on.Predicting or analyzing the traffic data is an important method to plan the city.Intelligent Transportation System(ITS)is a set of applications aimed at providing innovative services relating to different modes of transport and traffic management and enables users to be better informed and make use of more coordinated and smarter use of transport services.And with the popularity of more and more automotive equipment,traffic data collection becomes easier and data's attributes become more abundant and the quantity becomes larger.So,the research of traffic data is more reliable.This paper is to use the real Shanghai's taxi GPS data to predict the destinations,and the cellular signaling data in Changchun to study the location privacy protection.Nowadays,based on the importance of location-based services,taxis are an important way for users to travel.In this paper the proposed algorithm-Prediction algorithm based on time(PBT algorithm),which is considering the influence of time factor for destination prediction and is also based on Sub-Trajectory Synthesis algorithm.Experiments based on real datasets show that in terms of destination prediction,PBT algorithm has higher effectiveness than Sub-Trajectory Synthesis algorithm.At the same time,it is necessary to effectively protect the real location information of users in location-based services,so,this paper puts forward the location privacy protection algorithm.They not only achieve the location privacy protection in destination prediction,but also make the users' trajectory data to anonymize,so as to achieve the purpose of privacy protection.Therefore,the research on trajectory data has practical significance and important value for traffic intelligence.Main contributions: 1.The influence of time factor is proposed for predicting destination,based on the trajectory of the forecast period for destination,makes the prediction results more accurate.2.Model modeling.According to the trajectory of the destination in the different period of time.Markov model is different in the different period,so in different time period Markov model is established,and the state transition probability matrix is also different.Last constructing the three-dimensional state transition probability matrix where the Y axis is time and applied in the later steps of the algorithm.3.Use the mobile phone signaling trajectory data for privacy protection.In this paper,based on the map grid,and the location-based clustering algorithm in space,the user's travel path is anonymized to achieve the privacy protection.4.Use the real taxi GPS trajectory data and mobile phone signaling data for experiment,the experimental results proved that the destination prediction algorithm based on time and based on grid and trajectory clustering algorithm for the privacy protection are effective.