Data-Driven Cell Zooming for Large-Scale Mobile Networks

In large-scale mobile networks, energy minimization problem pertaining to base station (BS) switching is known to require high computational complexity. In this paper, we propose a data-driven energy-saving (DDE) framework which partitions the networks into communities. Each community can separately make decisions with low computational cost. We introduce a novel metric to measure the collaborative relationship among BSs. Then, the mobile networks are partitioned via a multiscale community detection method. To improve the cell zooming performance, we estimate the aggregate traffic demands of the communities based on the historical traffic profile. A heuristic switch-off strategy is proposed to maximize the energy savings while guaranteeing minimal service requirements. Experiments with two real regional cellular network datasets show that the proposed DDE framework can conserve a significant amount of energy and leverage the tradeoff between energy savings and the blockage ratio.