Hybrid Transceiver Design for Tera-Hertz MIMO Systems Relying on Bayesian Learning Aided Sparse Channel Estimation
Hybrid transceiver design in multiple-input multiple-output (MIMO) Tera-Hertz (THz) systems relying on sparse channel state information (CSI) estimation techniques is conceived. To begin with, a practical MIMO channel model is developed for the THz band that incorporates its molecular absorption and reflection losses, as well as its non-line-of-sight (NLoS) rays associated with its diffused components. Subsequently, a novel CSI estimation model is derived by exploiting the angular-sparsity of the THz MIMO channel. Then an orthogonal matching pursuit (OMP)-based framework is conceived, followed by designing a sophisticated Bayesian learning (BL)-based approach for efficient estimation of the sparse THz MIMO channel. The Bayesian Cramer-Rao Lower Bound (BCRLB) is also determined for benchmarking the performance of the CSI estimation techniques developed. Finally, an optimal hybrid transmit precoder and receiver combiner pair is designed, which directly relies on the beamspace domain CSI estimates and only requires limited feedback. Finally, simulation results are provided for quantifying the improved mean square error (MSE), spectral-efficiency (SE) and bit-error rate (BER) performance for transmission on practical THz MIMO channel obtained from the HIgh resolution TRANsmission (HITRAN)-database.