Efficient User Selection Algorithms for Wireless Broadcast Channels

This paper considers wireless broadcast systems with multi-antenna base stations. Preceding at the transmitter reduces interference between users allowing independent data streams to be sent to multiple users simultaneously. With typically more users than transmit antennas, efficient selection of user subsets is important. The paper compares the effectiveness of several user selection algorithms and identifies situations where they are suboptimal. The impact of zero-forcing (ZF) and dirty paper coding (DPC) precoding are considered. A modification to two current algorithms is proposed that reduces computational time while retaining effectiveness. Simulation is used to assess the modifications and suggest methods of setting parameters. The paper examines the variability of allocated data rates and shows how it affects the performance of weighted alpha rule schedulers, which trade-off efficiency with fairness. It is shown that reducing the variability of allocated rates can have a significant reduction on mean scheduler queue sizes.