Offset-Based Beamforming: A New Approach to Robust Downlink Transmission

Abstract

This thesis describes the design of low-complexity robust linear beamforming algorithms for multi-user downlink multiple-input single-output (MISO) communication systems. The goal of the algorithms is to provide the receivers with specified signal-to-interference-and-noise ratios (SINRs) with high probability under certain power constraints. Unfortunately, the SINR outage constraint is intractable, and precise formulations of these problems are fundamentally hard to solve. The contribution of this thesis is a suite of algorithms that provide high-quality approximate solutions to a broad range of robust downlink beamforming problems, and do so at low computational cost. The unifying feature of these algorithms is that they are based, either explicitly or implicitly, on a transformation of each SINR outage constraint into a non-negativity constraint on a random variable, and the approximation of that non-negativity constraint by offsets on the mean of the distribution. The first algorithm is developed for frequency division duplexing systems. Using a new extension of the S-Lemma, the channel uncertainty model is incorporated into the design problem using a zero-outage region approach. From that formulation, a new algorithm that is able to balance between the performance inside and outside the zero-outage region is developed. The resulting offset maximization algorithm has a low-complexity iterative closed-form solution that provides significant performance improvement, and can be extended to time division duplexing systems. Analysis of the offset structure reveals a refined notion of the offset that incorporates information about each user's channel, and results in a convex semidefinite relaxation problem. When the channel uncertainty size is small, further approximations lead to an approximate iterative closed-form solution. When the beamforming directions are defined in advance, that algorithm provides near-optimal power loading. Using subgradient methods, variants of the offset maximization algorithms that can accommodate per-antenna power constraints (PAPCs) are developed. Furthermore, the resulting offset-based power loading method can be combined with the maximum ratio transmission (MRT) or zero-forcing (ZF) directions, to provide robust algorithms that satisfy PAPCs with complexities low enough for massive MIMO applications. Finally, the principles of the offset maximization algorithm are applied to multi-cell systems with the centralized cooperation, and with the centralized and decentralized architectures. The resulting algorithms provide significant performance improvement over those existing in the literature, and do so at substantially lower computational cost.