FDD Massive MIMO: Efficient Downlink Probing and Uplink Feedback via\n Active Channel Sparsification

Abstract

In this paper, we propose a novel method for efficient implementation of a\nmassive Multiple-Input Multiple-Output (massive MIMO) system with Frequency\nDivision Duplexing (FDD) operation. Our main objective is to reduce the large\noverhead incurred by Downlink (DL) common training and Uplink (UL) feedback\nneeded to obtain channel state information (CSI) at the base station. Our\nproposed scheme relies on the fact that the underlying angular distribution of\na channel vector, also known as the angular scattering function, is a\nfrequency-invariant entity yielding a UL-DL reciprocity and has a limited\nangular support. We estimate this support from UL CSI and interpolate it to\nobtain the corresponding angular support of the DL channel. Finally we exploit\nthe estimated support of the DL channel of all the users to design an efficient\nchannel probing and feedback scheme that maximizes the total spectral\nefficiency of the system. Our method is different from the existing\ncompressed-sensing (CS) based techniques in the literature. Using support\ninformation helps reduce the feedback overhead from O(s*log M) in CS techniques\nto O(s) in our proposed method, with $s$ and $M$ being sparsity order of the\nchannel vectors and the number of base station antennas, respectively.\nFurthermore, in order to control the channel sparsity and therefore the DL\ncommon training and UL feedback overhead, we introduce the novel concept of\nactive channel sparsification. In brief, when the fixed pilot dimension is less\nthan the required amount for reliable channel estimation, we introduce a\npre-beamforming matrix that artificially reduces the effective channel\ndimension of each user to be not larger than the DL pilot dimension, while\nmaximizing both the number of served users and the number of probed angles. We\nprovide numerical experiments to compare our method with the state-of-the-art\nCS technique.\n