Widely Linear Processing Improves the Throughput of Nonorthogonal User Access

The quality-of-service (QoS) provided by wireless communication networks can be upgraded by the technique of non-orthogonal multiple-access (NOMA) in tandem with widely linear beamforming (WLB), which uses a pair of beamformers for each information symbol. Conventionally, rate-fairness among the users is achieved by maximizing the users’ minimal throughput (max-min throughput optimization). However, this is computationally challenging, as each iteration requires solving a high-dimensional convex optimization problem, even for small networks. We circumvent this by maximizing the geometric mean (GM) of the users’ throughput (GM-throughput maximization) and design novel algorithms based on iterating closed-form expressions are developed, which are shown to be hundreds of times more computationally efficient than the existing algorithms that are based on convex-solvers. The proposed algorithms are developed for both conventional wireless networks and networks requiring ultra-reliable and low-latency communications (URLLC).