Energy Efficient Resource Allocation for Hybrid Services with Future Channel Gains

TL;DR

This paper introduces a framework for enhancing energy efficiency in hybrid service systems by leveraging future channel gain predictions, demonstrating significant gains through joint resource optimization and a practical heuristic approach.

Contribution

It proposes a two-timescale resource allocation policy using future average channel gain predictions and introduces a low-cost heuristic for practical implementation.

Findings

Optimal policy outperforms relevant benchmarks.

Heuristic policy nearly matches optimal performance with perfect prediction.

Coarse-grained prediction suffices for energy efficiency gains.

Abstract

In this paper, we propose a framework to maximize energy efficiency (EE) of a system supporting real-time (RT) and non-real-time services by exploiting future average channel gains of mobile users, which change in the timescale of seconds and are reported predictable within a minute-long time window. To demonstrate the potential of improving EE by jointly optimizing resource allocation for both services by harnessing both future average channel gains and current instantaneous channel gains, we optimize a two-timescale policy with perfect prediction, by taking orthogonal frequency division multiple access system serving RT and video-on-demand (VoD) users as an example. Considering that fine-grained prediction for every user is with high cost, we propose a heuristic policy that only needs to predict the median of average channel gains of VoD users. Simulation results show that the optimal policy outperforms relevant counterparts, indicating the necessity of the joint optimization for both services and for two timescales. Besides, the heuristic policy performs closely to the optimal policy with perfect prediction while becomes superior with large prediction errors. This suggests that the EE gain over non-predictive policies can be captured with coarse-grained prediction.