Downlink Channel Parameter Prediction Based on Stacking Regressor in FDD Massive MIMO Systems

Abstract

Considering massive multiple-input multiple-output (MIMO) applications in the sixth-generation (6G) mobile networks. Due to the different frequency of uplink (UL) and downlink (DL) channels in frequency division duplexing (FDD) systems, the reciprocity between the UL and DL wireless channels is not valid. As a result, pilots are required to be sent both by the base station (BS) and user equipment (UE) for estimating the double-directional channels, which consume more transmission and computational resources. In this paper, we propose a DL channel parameter prediction method based on stacking regressor for FDD massive MIMO systems. It has a second-time prediction process, which uses multiple base regressors prediction results as features and meta-regressor as a model to realize DL parameter prediction. It is able to predict multiple DL parameters including path loss (PL), delay spread (DS), and angular spread. Both the UL channel parameters and environment characteristics are chosen as features to predict DL parameters. Simulation results have shown that the proposed method provides higher prediction accuracy than single base regressors and the 3GPP TR 38.901 channel model.