GDoF Duality of Treating Interference as Noise and Application to Power Control for Multiuser Interference Channels

In this work, we revisit the duality of treating interference as noise (TIN) from the generalized degrees-of- freedom (GDoF) perspective. Specifically, in a general K-user interference channel where interference is treated as Gaussian noise, we demonstrate that for any given GDoF tuple and its power allocation, how to derive a power allocation that yields a dominant GDoF tuple in its reciprocal channel. This finding naturally leads to a fast iterative GDoF-duality-based power control algorithm. Numerical results show that the proposed algorithm achieves competitive sum-rate performance with sophisticated SINR-based power control scheme.