Research On User Mobility Patterns Mining Based On Time

There are some regularions of a lot of users' mobility in our world and these are called user mobility patterns that have important meaning in many fields such as layout and design of mobile communication network, mobility management, and service based on location management etc. If user mobility patterns are related with other data information, such as business information, we will find users' behavior patterns, and consumers could be provided with individuation sevices based on location and changeful time. With the rapid development and widely application of mobile communication technique and computer technology, it is porssible to tail and locate users, and a mass dynamic time-space data about users' mobility have been accumulated in the mobile computing environment. The mobility logs are just ours necessary.Now data mining technology become perfect more and more, and it has been an important tool to extract user mobility patterns. Data mining could find useful information from lots of noisy, half-baked or inconsistent data sets. Sequence pattern mining is a very important means of data mining which was first put forward in 1995 by R. Agrawal and R. Srikant. In recent years, many domestic and abroad researchers have explored lots of algorithms. The methods of sequence patterns mining can't be directly applied to user mobility patterns mining which are similar to sequence patterns mining but have features themselves. At present the researches about mining user mobility patterns have been worked largely, however, lots of them have not taken time into account completely but only considered time as base to form user mobility sequences. But in our real life, user mobility has relation to time closely and that is why we work in this thesis.In this thesis, data mining and sequence patterns mining are summarized firstly, then user mobility management in mobility computing environment and several algorithms of user mobility patterns mining are introduced. We will find some flaws by analyzing user mobility pattern without time, and bring forward a new method about preprocessing data and computing support. Finally, we will provide an algorithm to mine user mobility patterns based on time.According to study in theory, a system about user mobility patterns and predict based on time is designed, which we could predict the future location of users by space-time factor, and data sets are generated by Stanford University. The result of experiment is that it is more accurate to predict the location by user mobility patterns based on time.