Angular domain pilot design and channel estimation for FDD massive MIMO networks

The huge training overhead for obtaining channel state information (CSI) at the BS has been recognized as a major challenge in frequency division duplex (FDD) massive multiple-input multiple-output (MIMO) cellular networks. To solve this problem, we propose an angular domain pilot design and channel estimation scheme to reduce the required overhead by exploiting the angle domain channel sparsity. Specifically, we firstly propose the downlink dominant angular set estimation by utilizing the directional reciprocity of FDD channels, where an index calibration algorithm is introduced to handle the effect of different wavelengths in FDD systems. Then, two kinds of angular domain pilots design schemes, named complete orthogonal pilot design and partial orthogonal pilot design, together with their corresponding feedback frameworks are proposed for channel estimation. Simulation results demonstrate that our proposed angular domain pilot design and channel estimation scheme could provide good mean square error (MSE) performance with much reduced pilot overhead, and consequently achieve much larger downlink throughput in comparison to the conventional scheme adopted in LTE.