MIMO Full-duplex Networks with Limited Knowledge of the Relay State

Full-duplex (FD)-enabled relay networks represent a relevant solution to two critical needs of next-generation networks, namely, radio coverage extension and high spectral efficiency of wireless communications. Under practical conditions, however, the FD mode may not be the best operational setting for the relay; rather, operating in half-duplex may be more convenient when harsh channel conditions add up to self-interference. One of the fundamental challenges in the design of FD relay networks is thus how to determine the relay operational mode and the value of transmit power at both the relay and the data source, so that the achievable data rate is maximized as time varies. We address this problem in a two-hop, MIMO network, accounting for practical operational conditions in which the source is unaware of the symbols that the relay is transmitting. In light of the problem complexity, we also derive a lower-bound to the maximum achievable rate, which proves to be tight, especially for low-medium SNR values. We then tackle massive MIMO networks, and exploit our asymptotic analysis in the number of antennas to derive a low-complexity, yet highly efficient, operational mode and transmit power allocation scheme for a finite-size scenario.