Energy-Efficient Proactive Scheduling Policies for Finite-Buffer Regular Service Guarantees

In this work, we study the energy saving merits of proactive scheduling for downlink multimedia streaming towards a finite-buffer receiver under Rayleigh fading. Three different threshold-based proactive scheduling policies are proposed, each with a different threshold structure on the queue/channel state space. The first two policies consider a single channel gain threshold per queue state, either with a fixed or variable cache amount, whereas the last policy imposes a set of thresholds on the channel gain, fixed on all queue states. We consider a time-causal system in which only the current and previous environment states are known. Required transmit power/signal-to-noise ratio (SNR) for the proposed policies is analytically derived, and numerically benchmarked against a non-proactive (reactive) transmission upper bound, as well as to a non-causal genie-aided lower bound with fully-observable future channel values. Numerical results show that proactive scheduling could save more than 50% of the transmission energy on average.