Massive MIMO multi-pair two-way half-duplex AF FDD relaying: channel estimation

We consider two-way amplify-and-forward FDD massive multi-input-multi-output (MIMO) relaying system, where multiple half-duplex (HD) user-pairs exchange information via a shared HD relay equipped with massive MIMO. To design and study the performance of these systems, we require channel state information in uplink as well as in downlink. For the uplink of the system model, we use conventional low-complexity MMSE estimator for the channel estimation, wherein the number of orthogonal pilots required for uplink channel estimation are of the order of the number of users. However, in the massive MIMO relay systems, the downlink channel estimation becomes challenging, because of increased pilot overhead in proportion to the number of antennas at the relay and employing conventional channel estimation approaches will result in poor spectral efficiency. The massive MIMO channels exhibits sparse structure in the angular domain, due to limited scattering environment. We exploit this channel sparsity to reduce the pilot overhead by performing sparse Bayesian learning-based downlink channel estimation. Furthermore, practical massive MIMO systems are built with low-cost hardware components which makes the massive MIMO system prone to hardware impairments like phase and quantization errors. In this paper, we numerically analyze the effect of hardware impairments on the two-way multi-pair half-duplex massive MIMO relay systems.