Mobility Prediction based on a Temporal Markov Model with User Social Ties
PSN, mobility prediction
Due to the rapid development of social media networks as well as the recent advances on smartphone technologies, users stopped being only consumers and became data producers. This enabled a new form of sensing, engaging a variety of studies using Location-Based Social Media (LBSN) data, such as device-to-device communication (D2D) where the user location is required to make mobile data offloading. For instance, if several users are requesting the same content and are in same area, a user can download it and share with them by other low-range interfaces such as Wi-fi Direct and Bluetooh, avoiding multiple connections with the base station (BS). In this context, spatial and temporal information plays a key role for analyzing user behaviors, which is useful for mobility prediction. In this paper, We propose Temporal Markov Model with User Social Ties (TMMUS), a mobility prediction algorithm based on spatial, temporal and social collective characteristics of all users as well as the personal ones. It is a model that takes into account location transitions of users' friends in order to acquire a higher accuracy. We have conducted extensive analysis with a LBSN real dataset, and our model outperforms the standard 1st-order Markov Models.