Research On Uncertain Path Prediction Model Of Moving Objects

In recent years, with the development of mobile computing technology, mobile network shows a profound impact in people’s life. Mobile network connect space and network, makes the internet a combination of all kinds of elements of real life. Smart phone, GPS equipped vehicle, smart household electrical appliances and other devices can generate a large number of trajectory and location data. These data can not only describes the historical trajectory of moving objects, but also accurately reflect the features of moving objects. We found the value of trajectory data because of improvement of technique of big data process. Reasonable use of trajectory data brings the opportunity to Internet operators. Therefore, more and more researchers pay more attention on location-based services. Particularly, trajectory prediction of moving objects that has wide applications gradually becomes a research hotspot.This thesis focuses on large scale of trajectory data, analyzes the characteristic of trajectories of moving objects without using road data, and introduces the method of clustering features from trajectories. Trajectories can be transferred into sequence by using clustered features. In addition, based on frequent pattern mining, a novel prediction method call PrefixTP was introduced. In order to prove the effectiveness and efficiency of the proposed algorithm, extensive experiments are conduct and the experimental results shows that the accuracy of prediction is improved by 37.04% while using PrefixTP algorithm. Based on the proposed PrefixTP method, a trajectory prediction system which provides a friendly-to-use interface and can display the detailed process of trajectory prediction was developed.The major contributions of this thesis are given as follows:(1) Introduces the background and relative works of trajectory mining. Mine direction changed points with cluster method by analyzing trajectory feature extraction. Improve cluster method by bring in space encoding method and provide the method of getting nearest neighbor. Transfer trajectories into sequences by using clustered features.(2) Analyze the typical trajectory prediction algorithms and find the different between Markov chain based trajectory prediction method and frequent patterns based prediction method. Detailed introduces the PrefixSpan algorithm. Based on the above techniques, trajectory prediction algorithm call PrefixTP was proposed based on PrefixSpan, which aims to predict trajectory of moving objects accurately.(3) Experimental tests were conducted using a real-world trajectory dataset generated by over 33,000 taxis in a period of 3 months. Experiments verify the accuracy and time performance of the proposed trajectory prediction algorithm. Moving objects trajectory prediction system is developed based on PrefixTP algorithm by using java programming language.