Capacity Scaling of Single-Source Multiantenna Wireless Networks Without CSIT

Abstract

We consider a wireless network in which a single-source node having m antennas transmits independent messages to n destination nodes, where each destination node has a single antenna. We assume that the source is located at the center of a unit area and the destinations are located uniformly at random in the same area. By applying transmit beamforming at the source, an achievable sum rate can scale as min {m,n}, which is defined by the aggregate rate of all messages. For transmit beamforming, however, channel state information (CSI) is essentially required at the transmitter (CSIT), which is hard to acquire in practice because of the time-varying nature of wireless channels and feedback overhead. We show that, even without CSIT, almost the same sum rate scaling law assuming CSIT is achievable by inducing cooperation between destinations. Specifically, we study the sum rate scaling law when both the number m of the source antennas and the number n of destinations increase such that n = m(beta) for beta > 0. If beta > 1 the optimal sum rate scales as m log m and if 0 < beta < 1 the optimal sum rate scales between m(beta(1 - epsilon)) and m(beta) log m, where epsilon > 0 is an arbitrarily small constant, which shows significant improvement compared to the case of no cooperation between destinations. Our result is of particular interest because we did not assume any additional bandwidth for cooperation.