Low-latency Low-complexity Subspace Methods for mmWave MIMO-OFDM Channel Estimation

Millimeter wave (mmWave) wideband channels in a multiple-input multiple-output (MIMO) transmission are described by a sparse set of impulse responses in the angle-delay, or space-time (ST), domain. In this paper we consider the problem of channel estimation and we discuss subspace methods which exploit the low-rank (LR) algebraic structure of the MIMO channel matrix and the related slowly- and fast-varying features (angles/delays of arrival and fading amplitudes, respectively). The main drawback of the optimal LR method is the excessively slow convergence to the mean square error lower bound for invariant angles/delay and time-varying fading. In this paper, new suboptimal LR techniques are proposed to reduce the complexity and accelerate the convergence. Numerical results show that the proposed methods closely approach the asymptotic bound with a number of slots that is two order of magnitudes lower than the optimal method, providing significant performance gains in realistic mmWave propagation scenarios.