Linear precoding for MIMO wireless channels

Abstract

Multiple input and multiple output (MIMO) technology is a powerful solution to high data rate and reliable communication on dispersive wireless channels. In close-loop MIMO systems where the channel state information is available at the transmitter(CSIT), it is possible to improve the system performance by optimizing a linear precoder based on certain objective functions and constraints. This thesis presents results we obtained on the linear precoder optimization for MIMO channels under different types of channel fading and system configurations. First, for MIMO flat fading channels with covariance feedback, we derive the optimal linear precoder which minimizes the pairwise codeword error probability upper bound of general Full-Rate Full-Diversity (FRFD) STBCs. Also, we find the existence of multiple minimum codeword different matrices (mCDMs) in real-rotation based FRFD STBCs, and thus formulate a joint optimization problem to take the worst effect of these mCDMs into account. Second, for MIMO frequency selective fading channels, we characterize the optimal linear precoder for time domain equalization (TDE) system in the sense of mean square error minimization. Interestingly, the optimal system so obtained is essentially a MIMO-OFDM system. Based on this observation, a low-complexity suboptimal scheme is proposed for efficient implementation. Finally, as MIMO-OFDM systems suffer from inherent high peak-to-average power ratio (PAPR), we then consider the linear precoder optimization for MIMO frequency domain equalization (FDE) system. It is shown that the proposed linear precoded FDE system enjoys similar error performance while maintaining much lower complexity and PAPR compared to the linear precoded TDE system, making it a very promising scheme for the broadband wireless uplinks.