Energy-efficiency and future knowledge tradeoff in small cells prediction-based strategies

Predictive small cells networks and proactive resource allocation are considered as one of the key mechanisms for increasing the long-term energy-efficiency of communication networks. Learning techniques exploit repetitive patterns in human behavior to predict some future transmission contexts of the network. In this paper, we target to improve the energy efficiency of delay-tolerant transmissions by enabling flexibility in resource allocation with prediction-based strategies. We study the performance, in terms of energy efficiency of several scenarios of future knowledge ranging from zero to perfect knowledge of the future context, but also partial knowledge scenarios (short-term predictions, long-term statistics or partial knowledge). An iterative process, approaching the optimal strategies in each scenario, is described. In some cases, closed-form expressions of the optimal strategies to be implemented can be obtained and the performance in each scenario is computed. Our analytical and numerical results assess the potential benefit of exploiting the knowledge of the future in the case of a delay-tolerant transmission and show how the system may benefit from a provided piece of information about the future transmission context.