Short-Term Power Allocation for Slowly Fading Channels Based on Markov Prediction

An optimal power allocation algorithm for delay-limited energy-constrained data transmission in block-fading channels, where blocks are separated in time, is proposed in this paper. Using this algorithm together with a discrete predictive model for the channel, an optimal power allocation strategy is obtained which minimizes the information outage probability based on the current channel state as well as a probabilistic knowledge of the future channel states. It is shown that for schemes with various orders of predictability of the channel, this algorithm outperforms equal power allocation in terms of the outage probability and the average maximum achievable rate.