Multiuser MIMO Systems with Random Transmit Beamforming

This paper considers the wireless downlink transmissions in a single cell environment, for which the base station (BS) is assumed to schedule its transmission to each mobile station (MS) on a time-slot basis. Only one MS is selected for transmission during each time-slot and the selected MS possibly changes from one time-slot to another. This transmission scheme is thus referred to as dynamic time-division multiple-access (D-TDMA). Random transmit beamforming with the feedback of effective signal-to-noise ratio (ESNR) was proposed by Viswanath and Tse IEEE Transactions on Information Theory, Vol. 48, No. 6, pp. 1277–1294, 2002] for D-TDMA-based systems in which multiple transmit antennas are equipped at the BS but only single receive antenna is equipped at each MS, or the so-called “MISO” systems. It was also shown in Viswanath and Tse, 2002] that when the number of MSs in the system becomes large, the system throughput achieved by random transmit beamforming converges to that by coherent transmit beamforming which, however, requires the complete channel state information (CSI) of each MS at the BS. This paper extends upon the work in Viswanath and Tse, 2002] to a more general scenario for which multiple transmit antennas and multiple receive antennas are equipped at the BS and each MS, respectively, or the so-called “MIMO” systems. We also consider several linear and nonlinear receiver structures and propose novel power allocation schemes to further improve the achievable system throughput. The throughput performance of the proposed receivers and power allocations schemes is compared through computer simulations and their fast convergence to the system throughput by coherent transmit beamforming is demonstrated.

Ying-Chang LiangEmail: