Scheduling and Precoding in Multi-User Multiple Antenna Time Division Duplex Systems

The downlink transmission in multi-user multiple antenna w ireless communication systems is generally studied assuming channel state knowledge and the topic of determini ng th s channel knowledge is considered as an unrelated topic. However, in practical interference-limited system s with mobile users, the two problems are tightly coupled, wi th a tradeoff existing between the two. In this paper, this coup ling is explicitly characterized as follows: channel train ing overhead and estimation error are rigorously accounted for while determining the net system throughput. First, a transmission method with training on reverse link only is co nsidered. Scheduling and precoding based transmission schemes are developed that effectively utilize the channel estimation process on the reverse link in improving net throughput. The schemes are applicable in the general setti ng of heterogeneous users with arbitrary weights assigned to these users, where the objective is to maximize net weight d-sum throughput. Next, a transmission method with forward link training in addition to reverse link channel tr aining is considered. In this setting, a different precodin g scheme is developed where the users utilize the forward pilo ts to estimate the effective channel gains. I. I NTRODUCTION The downlink and uplink transmission between a base station and a group of independent users in a multiple antenna setting as shown in Figure 1 is a complex problem with very many parameters that has received significant attention in recent years. The use of multiple antennas at th e ransmitter and receiver in a point-to-point communication system with the same power and bandwidth constrain ts has been shown to greatly improve the overall throughput of the system [1], [2]. This gain is due to the spat ial diversity obtained from the deployment of multiple antennas over a wireless medium. We are interested in the dow nlink transmission from the base station to the users. This multi-user multiple antenna downlink transmission sc enario is analyzed as the multi-antenna broadcast channel (BC) problem in information theory literature. The sum capa city of the multi-antenna Gaussian BC has been shown J. Jose and S. Vishwanath are with the Laboratory for Informa tics, Networks and Communications (LINC), Department of El ectrical and Computer Engineering, The University of Texas at Austin, Au stin, TX 78712 USA (email: jjose@ece.utexas.edu; sriram@e ce.utexas.edu). The authors were supported in part by NSF grants CNS-0626903 and CCF-0448181. A. Ashikhmin and P. Whiting are with Bell Laboratories, Alca tel-Lucent Inc., Murray Hill, NJ 07974 USA (email: aea@rese arch.bell-labs.com; pwhiting@research.bell-labs.com).