Order optimal energy efficient transmission policies in the small delay regime

In a wireless system with stochastically arriving traffic, scheduling the amount of data transmitted at any time is a basic technique for improving the energy efficiency. A discrete-time model for transmission scheduling over a fading channel is considered in this paper. The arrived data is placed into a transmission buffer, once the data is removed after the buffer; it is transmitted over the fading channel. The channel is modeled as a block-fading channel with Gaussian noise. For this model the optimal trade-off between the average queueing delay and the long-term average power was characterized in the asymptotic regime of large delays. The average power decreases as the average delay increases at the optimal rate. This rate can be achieved by a sequence of policies on the buffer occupancy via a simple threshold rule.