Energy-efficient link adaptation with transmitter CSI

Energy-efficient link adaptation is studied based on minimizing the total energy consumption per transmitted bit in a mobile terminal. It is shown that the optimal power allocation is water-filling; the optimal energy consumption per bit is a function of the power amplifier efficiency, the circuit power rate dependence, and a cutoff channel to noise ratio (CNR). For a given transceiver architecture, low energy consumption per bit corresponds to a high cutoff CNR and vice versa. The optimization of the energy consumed per bit is carried out for flat fading and frequency-selective fading channels. In the former, the optimization problem is solved analytically, resulting in optimal rate, power and energy consumption per bit being expressed in terms of the cutoff CNR. In the latter, the optimal power allocation is water-filling in frequency, depending on the subcarrier noise levels. Finally, highly efficient algorithms with superlinear convergence are proposed, which are based on the Dinkelbach method for solving nonlinear fractional programs.