Robust beamforming and base station activation for energy efficient downlink C-RAN

This paper considers an energy efficient (EE) maximization problem in a downlink cloud radio access network (CRAN), using the worst-case design criteria. We formulate the joint problem of robust EE and base station (BS) activation design by maximizing the worst-case EE under channel state information (CSI) uncertainty and per-BS power constraint, which is a nonconvex mixed-integer related fractional program and is difficult to solve. With the transformation of the worstcase signal-to-interference-plus-noise ratio (SINR), we propose a single-stage branch-and-bound (BnB) algorithm to find the global optimal solution via solving a sequence of second-order cone programming (SOCP) problems for any given active BS set. Then, a heuristic BS activation (BSA) algorithm is proposed to choose active BSS successively to further improve the worst-case EE. Simulation results suggest that the BnB algorithm converges to the global optimum, and a higher EE can be achieved by selecting the active BSS properly.