An Energy-Efficient Water-Filling Algorithm for OFDM Systems

In this paper, we develop an energy-efficient power allocation algorithm for the parallel channels of an OFDM system. This algorithm provides the optimum solution to a nonlinear fractional program involving an objective function called \textit{energy-per- goodbit} (EPG). The EPG objective function models the impact of both transmit power and constant circuit power consumption. The energy minimization problem formulation is quite general and subsumes both maximize rate (MR) and maximize margin (MM) problems as specific cases. As a result, the energy efficiency viewpoint provides a convenient and unified perspective of the various water-filling solutions. Using a numerical example, we show that the energy-efficient solution is quite different from the MM or MR solutions and can provide several dBs of performance improvement. We also study the impact of rounding to discrete constellation sizes.