Opportunistic Feedback and Online Optimization for Multiuser MIMO Systems with Linear Receivers

This paper proposes a scheduling and feedback strategy for the multiuser multiple input multiple output (MIMO) downlink. The base station uses a per-antenna scheduling that maps each transmit antenna to a different user in the downlink. The scheduling is performed based on the channel information obtained via a contention based feedback mechanism. Two design parameters, i.e., a channel quality threshold and a feedback probability, are proposed to adjust the scheduling and feedback behavior of this system. They are updated through a novel online updating procedure using empirically observed distribution information. The main advantages of the proposed protocol are that it substantially reduces the feedback requirements, still achieves multiuser diversity gain, and provides some fairness without a priori assumptions about the channel distribution.