NOMA has been proposed for downlink scenarios in 3rd generation
partnership project long-term evolution (3GPP-LTE)
systems [8]. In addition, NOMA has also been recognized
as a promising multiple access candidate for 5G wireless
systems [9], [10].
To further enhance the system capacity, the NOMA-based
multi-user beamforming has been introduced in [11] for
multiple-input single-output (MISO) NOMA systems. Unlike
conventional multi-user beamforming, NOMA-based multiuser
beamforming involves designing a single beamforming
vector to support multiple users. The number of supportable
users as well as the sum capacity can be increased. In [12],
a hybrid NOMA approach was proposed, where users were
grouped into small-size clusters, NOMA was implemented
within each cluster, and MIMO detection was used to cancel
inter-cluster interference. Subsequently, it was recognized
in [13] that the multi-user beamforming scheme proposed
in [11] may become inefficient when taking user fairness
into consideration. In [13], by adopting the idea of user
grouping, the so-called two-stage beamforming strategy has