Performance of MIMO spatial multiplexing algorithms using indoor channel measurements and models

Several algorithms have recently been proposed for multiplexing multiple users in multiple input multiple output (MIMO) downlinks. The ability of a transmitter to accomplish this using spatial methods is generally dependent on whether the users' channels are correlated. Up to this point, most of the multiplexing algorithms have been tested on uncorrelated Gaussian channels, a best‐case scenario. In this paper, we examine the performance of multiplexing multiple users under more realistic channel conditions by using indoor channel measurements and a statistical model for test cases. We use a block zero‐forcing algorithm to test performance at various user separation distances, optimizing for both maximum throughput under a power constraint and minimum power under a rate constraint. The results show that for the measured indoor environment (rich scattering, non‐line‐of‐sight), a separation of five wavelengths is enough to achieve close to the maximum available performance for two users. Since many spatial multiplexing algorithms require channel state information (CSI) at the transmitter, we also examine the performance loss due to CSI error. The results show that a user can move up to one‐half wavelength before the original channel measurement becomes unusable. Copyright © 2004 John Wiley & Sons, Ltd.