Approximate Message Passing for Downlink Sparse Channel Estimation in FDD Massive MIMO

An important challenge in massive multiple-input multiple-output (MIMO) systems operating with frequency-division duplexing (FDD) is to accurately estimate the channel response with low pilot signal overhead. Relying on the sparse structure of the channel vector, we investigate and apply three recently proposed algorithms based on Approximate Message Passing (AMP) to the channel estimation problem. As these algorithms are defined for real-valued problems, we propose a means to appropriately transform the complex-valued problem. In simulation experiments, the estimation error with reduced number of pilot signals and its dependence on signal-to-noise ratio (SNR) are evaluated. The effects of pilot matrix design and prior probability distribution modeling are also considered.