Battery-Aware Optimization of Green Small Cells: Sizing and Energy Management

Abstract

Motivated by the concerns about climate change, energy-harvesting technologies have recently gained particular interest to support the growing power demand of cellular networks. The environment-friendly power sources coupled with batteries enable to significantly reduce carbon emissions as well as the electricity bill of mobile network operators (MNOs). However, incautious battery usages can heavily damage the storage capabilities and require large expenses for the battery replacement. To investigate this tradeoff, we propose an energy management framework for a small cell powered by renewable energy, a battery, and the smart grid. First, we develop an approach that accounts for the battery aging to determine the optimal system sizing. Then, we design an energy controller, based on reinforcement-learning, which supervises the battery state in order to minimize the electricity expenditures of the MNO while enhancing the battery life span. Simulations show that the proposed solution achieves considerable cost reduction compared to a classical Kalman filter-based method proposed in the literature and performs very closely to the ideal strategy able to perfectly predict the state of the stochastic variables. Moreover, simulation results indicate that 30% of the battery life span can be saved each year by implementing the proposed solution.