Hybrid Message Passing Algorithm for Downlink FDD Massive MIMO-OFDM Channel Estimation

The design of message passing algorithms on factor graphs has been proven to be an effective manner to implement channel estimation in wireless communication systems. In Bayesian approaches, a prior probability model that accurately matches the channel characteristics can effectively improve estimation performance. In this work, we study the channel estimation problem in a frequency division duplexing (FDD) downlink massive multiple-input multiple-output (MIMO) orthogonal frequency division multiplexing (OFDM) system. As the prior probability, we propose the Markov chain two-state Gaussian mixture with large variance difference (TSGM-LVD) model to exploit the structured sparsity in the angle-frequency domain of the massive MIMO-OFDM channel. In addition, we present a new method to derive the hybrid message passing (HMP) rule, which can calculate the message with mixed linear and non-linear model. To the best of the authors' knowledge, we are the first to apply the HMP rule to practical communication systems, designing the HMP-TSGM-LVD algorithm under the structured turbo-compressed sensing (STCS) framework. Simulation results demonstrate that the proposed HMP-TSGM-LVD algorithm converges faster and outperforms its counterparts under a wide range of simulation settings.