Low-Rank Channel Estimation for Millimeter Wave and Terahertz Hybrid MIMO Systems

Massive multiple-input multiple-output (MIMO) is one of the fundamental technologies for 5G and beyond. The increased number of antenna elements at both the transmitter and the receiver translates into a large-dimension channel matrix. In addition, the power requirements for the massive MIMO systems are high, especially when fully digital transceivers are deployed. To address this challenge, hybrid analog-digital transceivers are considered a viable alternative. However, for hybrid systems, the number of observations during each channel use is reduced. The high dimensions of the channel matrix and the reduced number of observations make the channel estimation task challenging. Thus, channel estimation may require increased training overhead and higher computational complexity. The need for high data rates is increasing rapidly, forcing a shift of wireless communication towards higher frequency bands such as millimeter Wave (mmWave) and terahertz (THz). The wireless channel at these bands is comprised of only a few dominant paths. This makes the channel sparse in the angular domain and the resulting channel matrix has a low rank. This thesis aims to provide channel estimation solutions benefiting from the low rankness and sparse nature of the channel. The motivation behind this thesis is to offer a desirable trade-off between training overhead and computational complexity while providing a desirable estimate of the channel.