Optimality of treating interference as noise in the IRC: A GDoF perspective

Abstract

Interference is regarded as one of the major limitations in achieving the high date rates as required by the next generation of wireless networks (5G). One way of dealing with interference is using a relay. For instance, the deployment of a relay in an interference channel (IC) can improve the rate performance. However, this might require complicated decoding processes at the receivers in general. A question which arises is whether high complexity at the receivers can be avoided without loss of optimality. This motivates the study of treating interference as noise (TIN) in the interference relay channel (IRC). In particular, we characterize regimes in which TIN is optimal for the IRC in terms of generalized degrees of freedom (GDoF). To do this, two new upper bounds on the capacity of the IRC are established. In addition to some cases in the weak interference regimes, it is shown that TIN is optimal in the strong interference regime as long as the transmitter-relay link is sufficiently stronger than the relay-receiver link. The results indicate that computational complexity can be moved to the infrastructure, while the mobile devices can be kept simple and cheap without loss of optimality.