Adaptive precoding for wireless MIMO broadcast channels with limited feedback

Abstract

Recently, a number of techniques have been introduced to exploit multiuser diversity of a wireless multiple-input multiple-output (MIMO) broadcast channel (BC) that consists of a base station with $t$ transmit antennas and $K$ users with multiple antennas. However, prior works have ignored the rate overhead associated with feedback of MIMO BC channel state information at transmitter (CSIT), which is roughly $K$ times larger than single-user MIMO CSIT (i.e., it is $\mathcal{O}(tr)$ where $r = \sum_{k=1}^K r_k$ and $r_k$ is the number of antennas at the $k$th user). Considering the amount of feedback signaling, quantization is a necessity for effective feedback transmission as a form of partial CSIT. In this paper, we propose the greedy multi-channel selection diversity (greedy MCSD) scheme based on \emph{block} MMSE QR decomposition with dirty paper coding (block MMSE-DP), where partial CSIT is almost sufficient. The sum-rate performance of our novel scheme approaches extremely close to the sum capacity of MIMO BC as the number of users increases, whereas the feedback overhead is reduced by a factor of $2t^3/L(t^2-t)$, in which $L$ is the number of \emph{active} channel vectors. Simulation results validate the expectation from the analysis. In addition, the proposed scheme is shown to be appropriate for reconfigurable implementation.