Position-Time Pattern Based Method for Analyzing Users' Mobility.

Mobile network users’ movements analysis has increasingly importance with the mobile networks evolution. Time-spatial information about current and previous locations can contribute in optimizing mobile network performance. Extracting features from this information is the most common method in analyzing users’ mobility. In this paper, a new position-time pattern based method is proposed to analyze the mobility by translating time-spatial records of each user to target-oriented image. Dataset provided from an ISP in Shanghai city for one month is used in our work. The proposed method is deducing user’ mobility behavior by visualizing serving base stations positions vs time according to unified rules for all users. Mobility pattern ambiguity resulting from neighboring base stations is processed. Moreover, K-Nearest Neighbors based method is suggested to predict base stations neighborhood. Results shows that visualizing time-spatial dataset, with proper processing of neighboring base stations issue, provide better understanding of users’ mobility data analysis.