A two-stage rank selection scheme in downlink CoMP transmission networks

This paper considers a downlink Coordinated Multi Point (CoMP) transmission network, where the base stations (BSs) are connected to a central unit via backhaul links to serve mobile users (MUs). The objective is to jointly optimize the BS mode (active or sleep), MU association, and cooperative beamforming to minimize the total power consumption including backhauling power. We formulate the problem as a mixed-integer non-convex problem under the constraints of signal-to-interference-plus-noise ratio (SINR) requirements and limited BS power. By exploiting a hidden convexity structure of the SINR, the problem is first transformed equivalently into a mixed-integer second-order core programming (MI-SOCP) problem. Due to the coupled integer variables, the complexity of exhaustive search grows exponentially with network size, which motivates developing low-complexity heuristic algorithms. We propose a novel Two-stage Rank Selection (TRS) method to approximately solve the coupled MI-SOCP problem by determining the BS operation mode and MU association successively, which enables the remaining beamforming computation to be achieved through an efficient SOCP solution. We compare the computational complexity and numerical performance of TRS against several state-of-the-art alternatives. Analytical and simulation results demonstrate that the proposed method can significantly reduce the computational complexity and the total power consumption over existing alternatives.