We study a relay-aided downlink cellular network with one
NB-antenna BS, L MIMO FDRs, and L single-antenna users,
as illustrated in Fig. 1. Each FDR is equipped with Nt transmit
antennas and Nr receive antennas, and performs decode-andforward
transmission to avoid noise/interference amplification.
IRI is considered between any two relays. Similar to [18] [19],
we assume that the direct links between the BS and the users
are ignored due to severe path loss and blockage. Typically,
the users are dispersed geographically in a cell and associated
with different relays. While more than one user may exist in
each relay’s coverage, we consider a simplified model where
each relay selects only one user to serve1: the data intended
for user i is first received and decoded by relay i on feeder
link i, and then re-encoded and forwarded to user i on access
link i. To account for the case that users can possibly be
located at the coverage overlap of adjacent relays, MUI is
considered for each user from relays other than its serving
relay. This model can be prospectively extended to the more
general scenario with multiple users served by each relay in a
multi-cell network, which is left for future works.
On each feeder link, multi-stream beamforming2 is applied
at BS. Assuming linear beamformers, the received signal at
1Our model can be extended to the scenario for a relay to serve multiple
users, and scheduling is also necessary to improve the system performance.
Anyway, selecting a “good” user for each relay, e.g. a user with strong access
link, is a simple yet efficient scheme.
2Also known as “precoding” in some literatures. For simplicity, we use the
term “beamforming” regardless of single-stream or multi-stream transmission
in this paper