User scheduling and grouping in massive MIMO broadcast channels with heterogeneous users

摘要

This paper considers a frequency division duplexing massive multi-input multi-output system with two-stage beamforming. In two-stage beamforming, user scheduling and grouping are applied to reduce inter-user and inter-group interferences, respectively. Conventional scheduling utilizes the channel norm as a user selection metric. However, the metric degrades the multiuser diversity gain due to inter-group interference. To increase multiuser diversity gain from scheduling, an approximate lower bound on expected signal-to-interference-plus-noise ratio (SINR) value is derived in this paper. Its asymptotic sum rate is also derived in a realistic channel model where each user is randomly located in the cell. Analysis results show that the sum-rate growth increases with the pathloss exponent, but is upper-bounded due to inter-group interference. In addition, user grouping for randomly-distributed users is proposed. The approximate lower bound in user selection is used to obtain a new grouping metric by averaging over an instantaneous channel. In simulations, the proposed metric provided higher sum rate than did conventional user grouping methods.