by Time-Varying Beamforming in Multiuser Systems with Slowly Varying Fading Channels

We analyze a cellular communication system in which a basestation or access point equipped with several antennas transmits packet data to several mobile data users by means of a TDMA scheme. A scheduler in the basestation uses knowledge about the instantaneous channel quality (signal to noise ratio, SNR) of all users to allocate the channel resource to the user which maximizes a certain metric. By this means the system exploits multiuser diversity, which increases system throughput (with constant Quality of Service) particularly if the rate of variation of the users' SNR is fast. In cases when the underlying physical channel dynamic is insufficient, time-varying beamforming is employed in order to increase the effective channel dynamic. We analyze the implication of time-varying beamforming on coherent channel estimation, which is performed by all users for coherent detection, and discuss fundamental tradeoffs between system (and user) throughput and estimation performance. I. INTRODUCTION Wireless data traffic is forecasted to be increasingly asym- metrical, which is why the downlink of future cellular systems will constitute the bottleneck and requires optimization. The efforts towards meeting future demands for high rate wireless data must be focussed on increasing the spectral efficiency without increasing either transmission bandwidth or transmit power. This can only be achieved by systematically utilizing all available forms of diversity. Presently, diversity forms such as spatial, multipath or code diversity aim at either increasing the quality of a single communication link, or increasing the number of terminals that can communicate with a basestation or access point. In an environment where one transmitter communicates with many mobile receivers, another form of diversity can be exploited in order to improve spectral efficiency: multiuser diversity. If the transmitter has knowledge about each receiver's channel quality and a packet- based transmission protocol is used, a "smart" channel aware scheduling algorithm transmits data packets only to users with "good" channel conditions. The gain obtained by scheduling users in such a way is large when fading for all users is fast and has a large dynamic range. In pico- or microcells with typically slow fading it can be shown that employing multiple antennas along with time-varying (or random) beamforming increases multiuser diversity by "randomizing" the channels (1). This is a completely new design paradigm, which is in contrast to conventional approaches where one attempts to exploit diversity in order to average over the fading process. In this paper, we analyze the effect of time-varying beamforming on coherent channel estimation performance in a system where this technique is used to increase the effective fading rate of the channel. We assume that all users perform data-aided linear mimimum mean squared error (LMMSE) channel estimation and we assess the implication of increasing the channel dynamic on estimation performance. The paper is organized as follows: the system model is presented in section II, the LMMSE channel estimation and modifications thereof for the underlying effective channel are discussed in section III. Section IV presents system simulation results before a conclusion and an outlook to future work is given in section V.