Linear precoding in large scale MIMO under 3D channel model

Large scale MIMO is a new field for wireless communication. Theoretical analysis was usually done with independent distributed complex Gaussian channels and unlimited antennas. In this work, we exhibit the performance of linear precoding under the three dimension (3D) channel model calibrated by 3GPP in the 3D-UMÌ scenario. Considering the multi-cell multiuser MIMO (MU-MIMO), we assume that the base station (BS) has perfect channel state information (CSI) due to the reciprocity of TDD system. The BS is equipped with two dimension (2D) antenna array and the transmitted signal will experiences 3D radio propagation. We compare two linear precoding, maximum ratio transmission (MRT) and zero forcing (ZF), when the number of antennas is 16, 32 64 and 128. Also the polynomial expansion linear detector (PELD) method here is leveraged. Simulation results show that under the 3D channel model, as the number of antennas goes larger, the performance of both linear precoding goes better. Due to the limited number of antennas and the 3D propagation channel model, the inter-beam interferences deteriorate the performance of MRT. The performance of ZF is obviously better than MRT. Though the ZF precoding can eliminate the inter-beam interference, processing complexity is huge. The PELD precoding can achieve a suitable performance as a tradeoff between MRT and ZF.