Performance and Complexity of MIMO Detectors for Advanced Wireless Communications Systems

摘要

The multiple-input multiple-output (MIMO) concept has recently been pursued alongside orthogonal frequency-division multiplexing (OFDM) for advanced wireless communications systems. Our paper focuses on the performance and complexity of several MIMO OFDM detectors for frequency-selective Rician fading channel. Thus, we consider the zero-forcing (ZF), minimum mean square error (MMSE), and maximum likelihood (ML) detectors for the realistic channel models obtained from measurements by the WINNER II project. These models indicate that the fading is typically Rician and that the delay spread (DS), azimuth spread (AS) of the Laplacian power azimuth spectrum, and the Rician K-factor have scenario-dependent lognormal distributions and correlations. Previous MIMO detection performance and complexity evaluations have typically assumed Rayleigh fading and arbitrary DS, AS, and K values. Our paper shows numerical simulation results and complexity assessments for ZF, MMSE, and ML MIMO detection for Rician vs. Rayleigh fading as well as for random vs. fixed DS, AS, and K. The simulations have been done in MATLAB using the recently added function mimochan to generate the channel matrix samples. Previous MIMO evaluations have typically generated these samples using custom-built (nonstandard) code, which may be less reliable. Our results can more accurately predict MIMO OFDM performance in actual channels and should be useful to designers of advanced wireless communications systems such as WiMAX, WiFi, and LTE-A.