An Efficient User Selection Algorithm for Zero-Forcing Beamforming in Downlink Multiuser MIMO Systems

Abstract

SUMMARY In this letter, we propose an efficient user selection algo-rithm aiming to select users with less spatially correlation and meet the usernumber limit of zero-forcing beamforming in downlink multiuser MIMOsystems. This algorithm yields a considerable complexity reduction withonly a small loss in performance and it only needs partial users’ CSI feed-back. Coupled with the algorithm, a null space updating method in O ( K 2 )time and a modified proportional fair scheduling algorithm are also pro-posed. key words: user selection, zero-forcing beamforming, multiuser MIMO,null space updating 1. Introduction In recent years, there has been considerable interest in wire-less multiple-input, multiple-output (MIMO) communica-tions systems, due to their promising improvement in chan-nel capacity. It has been known that when channel state in-formation (CSI) is available at the transmitter in multiuserdownlink systems, the base station (BS) can transmit in-dependent data streams to multiple users simultaneouslyto achieve a much higher channel capacity than conven-tional MIMO techniques. Dirty Paper coding (DPC) [1] is apromising solution to this problem and it has been shown toachieve the sum-rate capacityof the multiple-antenna broad-cast channel [2]. However, DPC requires a new class ofcodes that are interference-dependent. The complexity ofthis algorithm makes it unsuitable for actual implementationat BS. Zero-forcing (ZF) beamforming is a sub-optimal butsimpler approach to this problem, where all inter-user inter-ference is driven to zero [3]. It is shown that a ZF beam-forming algorithm can attain the same asymptotic sum ca-pacity as DPC, as the number of users goes to infinity [4].In ZF beamforming strategy, the number of antennas in thebase station must be greater than the total number of anten-nas of all receiver users [3]. Selecting the best active userswith less spatial correlation can meet user number limit ofZF beamforming algorithm and at the same time with less