Hybrid precoding in Massive Antenna Multi-User System with Modified Hybrid Connected Structure

In this paper, a new hybrid precoding architecture is proposed for massive antenna multi-user system. The proposed architecture is an extension of the hybrid connected structure (HCS), which can realize a good tradeoff between spectral efficiency and hardware cost by adjusting the number of sub-arrays. Based on the proposed architecture, the hybrid precoding matrix is optimized by using the alternating optimization (AO) and Riemannian conjugate gradient (RCG) algorithm. The proposed scheme is evaluated based on real measured channels at 28GHz in a railway station. Simulation results show that the achievable sum rate of proposed scheme is very close to that of the optimal full-digital precoding scheme. In addition, the achievable sum rate of the proposed scheme is higher than that of the existing schemes when the number of sub-arrays takes different values.