Discrete adaptive transmission for fading channels

In this work, we address optimal adaptive transmission policies in slow varying wireless environments. Continuous rate and power assignments that achieve the ergodic capacity for these channels have been derived previously. Nevertheless, from a practical point of view, use of a finite number of power and rate levels is imperative. Here, we address the mapping from channel states of an arbitrary distribution to a discrete set of power level and code rate pairs. Unlike earlier work, our design does not require that the transmitter knows the exact value of the current channel state. We show that our design yields results close to the well-known water-filling result.