Precoding Weight Interpolation Method to Mitigate Spectrum Efficiency Loss of Vector Perturbation Based Multiuser MIMO with Limited CSI Feedback

摘要

Vector perturbation (VP) algorithm requires perfect channel state information (CSI) for optimal perturbation vector search and generating a spatial filter to pre-equalize the MIMO channel function. Due to the limited CSI feedback and realistic computational complexity (CC) due to hardware implementation, MU-MIMO OFDM system will yield the inefficient perturbated vector and mismatched spatial filter which causes large spectrum efficiency (SE) loss. In this paper, by simulation, it shows that most of performance loss of VP algorithm is derived from the mismatched spatial filter. Therefore, we propose a low-complexity linear interpolation for spatial filter to reduce the SE loss. The proposed interpolation scheme firstly rebuilds the CSI of partial subcarriers from limited feedback CSI by time and frequency domain transformation (TFT). After that, the spatial filter matrices of these partial subcarriers are calculated. Finally the spatial filter matrices of all subcarriers are interpolated by a linear interpolation. The complexity analysis and simulated results show that the proposed interpolation scheme can reduce the SE loss of VP based MU-MIMO OFDM system with low computational complexity.