Analysis and simulation of MIMO zero-forcing detection performance for correlated and estimated Rician-fading channel

This paper studies zero-forcing (ZF) detection for multiple-input multiple-output (MIMO) wireless communications systems and transmit-correlated and estimated Rician fading with realistic azimuth spread (AS) and K-factor. Due to the nonzero-mean channel gain matrix, the matrix that determines the symbol-detection signal-to-noise ratio for the spatially multiplexed streams has a noncentral Wishart distribution, which encumbers the derivation of an average error probability (AEP) expression. Therefore, we resort to an increasingly-popular approximation with a central Wishart distribution. We demonstrate, for the first time, that this approximation is reliably accurate only for rank-one deterministic (mean) channel matrix. Nevertheless, this approximation simplifies the derivation of a ZF AEP expression for transmit-correlated Rician fading and perfect CSI. For estimated CSI, we propose a MIMO ZF detector that employs both the channel gain estimates and, unlike conventional ZF, the channel statistics. The AEP for the new detector is derived. Numerical results from the AEP expression and from simulation are shown for realistic scenarios with Laplacian power azimuth spectrum. They reveal that the new detector outperforms the conventional detector. Furthermore, the AEP expression for the new ZF detector is found useful in evaluating the AS and K effects on performance.