Efficient Channel Tracking Based on Compressive Sensing for OFDM Millimeter-Wave Systems

In this paper, we propose a new channel tracking method to track the time fluctuations of millimeter-wave (mmWave) channels over multiple subcarriers with few training overheads and low complexity. In essence, channel tracking is formulated as a compressive sensing problem with the proposed sensing matrix over the delay domain, which enables low complexity irrespective of the number of available subcarriers, resulting in a new tracking algorithm based on the exact top-
$k$
feature selection. To elaborate, the proposed method is based on a novel sensing matrix that exploits the time evolution models of angle of arrivals (AoAs) and angle of departures (AoDs), where the time-varying channel is effectively approximated with a few discrete AoA and AoD candidates. Moreover, the proposed tracking algorithm is refined by introducing multiple sub-sensing subsets, which further improve the tracking performance. Numerical results and complexity analyses in terms of floating operations (FLOPs) confirm that the proposed approaches achieve better trade-offs between complexity and tracking gain than the conventional Bayesian estimator approaches.