Research And Implementation Of Mobile Location Prediction Method

With the rapid development of modern information technology, Data shows a trend of rapid growth. The analysis method of data has faced a serious challenge. Researching on the analysis method of data have value. For a long period of time in the past, human movement is considered to be chaotic behavior, random, no rules can be found. However, with the development of location technology and popularization of smart mobile devices, a large number of data which can reflect human movement behavior is accumulated. These data not only cover a wide range of people, but also reflect the feature of human mobile behavior from multiple angles. Predicting mobile location is an important problem in the research of human mobile behavior. It has an important significance to the design of mobile communication network and mobility management.This paper summarizes the basic theory and current research status of human mobility and mobile location prediction. The existing prediction algorithm of mobile location based on Markov process has the problem of low efficiency and data mutual interference. We propose a kind of improved Markov mobile location prediction method. This method firstly designs and implements a grid density clustering algorithm used to identify the user visiting location. It restrict visiting range by introducing grid and promote identification efficiency using density-based algorithm. Further design and implement a measurement algorithm of the similarity between the users in the movement behavior. By gathering user data according to their similarity, generating state transition matrix for each group of users. Both to solve the data sparsity problem when using a single user’s trajectory data to establish state transition matrix and to solve the problem of data’s mutual interference when using global user trajectory data to establish state transition matrix. Finally, this paper tested the method’s running time on different scale datasets, proving that grouping Markov mobile location prediction algorithm is superior to the traditional Markov mobile location prediction algorithm in efficiency.