Massive MIMO Channel Prediction: Kalman Filtering Vs. Machine Learning

This paper focuses on channel prediction techniques for massive multiple-input multiple-output (MIMO) systems. Previous predictors are based theoretical models, which would be deviated from realistic channels. In this paper, we develop and compare a vector Kalman filter (VKF)-based predictor machine learning (ML)-based using the channels spatial model (SCM), has been adopted in 3GPP standard years. First, propose low-complexity mobility estimator average large number of antennas MIMO. The estimate can used to determine complexity order developed predictors. VKF-based exploits autoregressive (AR) parameters estimated SCM Yule-Walker equations. Then, ML-based linear minimum mean square error (LMMSE)-based noise pre-processed data is developed. Numerical results reveal that both have substantial gain over outdated terms accuracy rate. larger overall computational than predictor, but once trained, operational becomes smaller predictor.