Path-Selective Precoding for FDD-based Massive MIMO Systems

The main purpose of this paper is to propose an effective precoding technique for the frequency-division-duplexing (FDD)-based massive MIMO systems under the common scattering effect. Key idea of the proposed path-selective precoding (PSP) scheme is to choose a small paths maximizing the data rate and then exploit only the angle information of the chosen paths for the downlink data precoding. To efficiently select the paths for each mobile, we use the notion of leakage, a metric of how much signal power leaks into other mobiles. While the interference is a joint function of precoding vectors of different mobiles, the leakage is solely a function of the precoding vector of corresponding mobile so that the signal-to-leakage-and-noise-ratio (SLNR) maximization problem can be decoupled into the sub-problems for each mobile. To find out a near-optimal solution of the decoupled SLNR maximization problem, we propose a greedy algorithm that iteratively removes the index of shared paths from the candidate index set until the SLNR does not increase. We demonstrate from the simulation results that the proposed PSP scheme achieves the significant data rate gains over the conventional angular-domain precoding schemes.