Abstract
With the population of mobile phones, the telecom company has a user's cellular phone signals and its movement trajectory. So, the company can learn about the user's activity habits and predict where the user may go next to meet the need of network resource and service management. In this paper, we propose a mobility prediction framework with Many-to-one Recurrent Neural Network (RNN). First, we extract the place that the user frequently visits (i.e., Points of Interest (POI)) from the user's mobility data through our proposed POI mining method, i.e., acceleration clustering. Then, we propose an adaptive mapping method to map the user's trajectory to a series of POI. Afterward, we use the RNN with Long Short-Term Memory to learn the user's POI series. Finally, we evaluate the prediction performance of the proposed scheme on two different real datasets. The performance shows that the prediction accuracy of our scheme outperforms previous works.