Compressive Channel Estimation in FDD Multi-Cell Massive MIMO Systems with Arbitrary Arrays

Abstract

Efficient downlink channel state information acquisition at the base station is crucial to achieve the potential gains of FDD massive MIMO systems. Conventional approaches for channel estimation require a training and feedback overhead which scales with the number of base station antennas, which make them unsuitable for large scale FDD MIMO systems. Alternative strategies exploiting the sparsity in the massive MIMO channel rely on additional assumptions on a shared common support between downlink channels, a specific array geometry at the base station (BS) and user equipments (UE), or a given modulation scheme. In this paper we propose a general approach for downlink channel estimation in FDD massive MIMO systems which leverages the individual sparsity in the different downlink channels without any other additional assumption. The designed precoding strategy allows channel estimation at the UE without knowledge of the BS array geometry. Simulation results show the accuracy of the estimations and the significant overhead reduction over conventional approaches.