Thresholding-based distributed power control for energy-efficient interference networks

In this paper, we propose simple one-shot power control functions and assess their performance both through analytical and numerical results. The proposed functions only assume individual channel state information (CSI) at each transmitter and are based on channel inversion and more importantly on thresholding; a transmitter uses zero power if the channel gain is below a threshold. Although the idea of thresholding has been used for maximizing spectral efficiency, it has not been used for maximizing (the total network) energy-efficiency (EE), which is measured here in terms of sum-EE. More specifically, we prove the optimality of the proposed policy in asymptotic regimes such as the low and high interference scenarios. We also prove that the expected sum-energy is individually quasi-concave with respect to each of the thresholds; this allows us to provide a low-complexity algorithm which can be run offline to find good thresholds. Through numerical simulations, we show that the simple idea of thresholding provides very appreciable gains.