Learning to Predict the Mobility of Users in Mobile mmWave Networks

MmWave communication suffers from severe path loss due to high frequency and is sensitive to blockages because of high penetration loss, especially in mobile communication scenarios. It highly depends on line-of-sight channels and narrow beams, and thus efficient beam tracking and beam alignment are necessary techniques to maintain robust communication links, in which tracking user mobility lays the foundation for beam tracking. In this article, ML techniques are applied to learn the mobility of the mobile mmWave users and predict their moving directions. Moreover, this article builds up an experiment environment by using the National Instruments mmWave transceiver system and our designed high gain antenna operated at 28 GHz carrier frequency, and then collects experimental data of the transmitted mmWave signals, which are next trained by deep learning algorithms. A deep neural network is learned and then used to predict a user's moving direction with up to 80 percent prediction accuracy in mmWave communication without the support of traditional channel estimation.