A Graph-Based ABS-Assisted TBS Sleep Scheme

Abstract

The large-scale deployment of 5G base stations has led to a significant increase in energy consumption. To enable green communication, we propose a graph-based aerial base stations (ABS)-assisted terrestrial base station (TBS) deep sleep scheme. In this scheme, TBSs in cells during low-traffic-demand periods turn off most of their communication devices and go to the sleep mode with less power consumption. As a substitute, the offloaded traffic demands are taken over by the on-demand deployed ABSs. When the predicted traffic in the cells is low, an ABS can be deployed in advance to cover and serve multiple cells. Thus, we jointly optimize the cell clustering and ABS deployment. Specifically, we first propose a convex optimization based search algorithm to obtain all feasible cell clusters. To reduce the computational complexity, a randomized incremental based search algorithm is designed for the homogeneous networks. Then, the primal problem is recast as the maximum weight independent set problem in graph theory, and an efficient algorithm is adopted to solve it. Numerical simulation results demonstrate that our algorithm is superior to the comparison schemes and significantly degrades the computation time.