Reduced Complexity Approximate Message Passing for Hybrid Architecture Based Millimeter Wave Massive MIMO Channel Estimation

Abstract:
We develop a low complexity channel estimation algorithm for millimeter wave massive MIMO systems deploying a hybrid architecture. Motivated by the sparsity of the millimeter wave massive MIMO channel in the angular domain, many compressed sensing-based channel estimation algorithms have been developed. As the number of antennas grows, the size of the sensing matrix becomes very large, leading to high complexity and large storage requirements. In this paper, we develop a low complexity Bayesian approximate message passing based channel estimation scheme. The complexity and the storage requirements will be reduced by designing analog and digital precoders that result in a sparse sensing matrix, which achieves a recovery performance close to a full-size random Gaussian sensing matrix.