Robust MISO Downlink: An Efficient Algorithm for Improved Beamforming Directions

The design of a set of beamformers for the multiple-input single-output (MISO) downlink that provides the scheduled users with the quality-of-service (QoS) level that they requested can be quite sensitive to the accuracy of the channel state information (CSI) that is available at the base station. To mitigate that sensitivity, models for the uncertainty in the CSI can be incorporated in the design formulation, but the resulting optimization problems are typically difficult to solve. A recently developed low-complexity algorithm for the power loading problem, in which the beamforming directions are pre-defined, has demonstrated good performance in numerical experiments. In this paper the reasons that underlie that good performance are examined. Then, using insights from that analysis, a computationally-efficient algorithm for jointly designing the beamforming directions and the power loading is developed. The resulting beamformers provide a significant reduction in the outage probability, and a hybrid of the two algorithms provides even further reduction.