Prediction of Human Mobility Using Mobile Traffic Dataset with HMM

Abstract

The connectivity and usability of cellular communications influencing to infer the information about user mobility based on the mobile traffic data. Thus, mobile traffic data can be used to analyze human location histories and prediction of human mobility. This paper proposes a framework for predicting human mobility which is based on Hidden Markov Models (HMMs). First, locations are clustered according to their characteristics such the highest traffic generated in a particular location, in certain time period. Then, the proposed HMM will be trained by the generated clusters. The usage of HMMs empowers to deal with spatiotemporal data, location characteristics, and possible visited states which are called the observable states.