Channel Estimation for Dynamic Metasurface Antennas-Based Millimeter Wave Massive MIMO

The amalgamation of millimeter wave (mmWave) and massive multiple-input multiple-output (MIMO) can significantly enhance system capacity and achievable rate, albeit at the cost of power or financial expenditure. Dynamic metasurface antennas (DMAs) emerges as a promising hardware- and powerefficient antenna paradigm for realizing large-scale antenna arrays systems. To fully enjoy its potential, we investigate the channel estimation problem in DMAs-based mmWave massive MIMO systems. Specifically, we propose a novel channel training pattern accounting for the array constraints of DMAs and adjustable characteristics of metamaterial elements. By leveraging the inherent sparse scattering properties of mmWave channels, we formulate the received training signal as a third-order tensor, whose factor matrices encapsulate the physical multipath parameters of the high-dimensional channel. Subsequently, we propose a tenser decomposition-based channel estimation algorithm by leveraging the Vandermonde-structure factor matrix, allowing to extract the multipath parameters of the channel from the decomposed factor matrices, and thereby facilitating the reconstruction of the entire massive MIMO channel matrix. In addition, we derive the Cramer-Rao lower bound for the DMAs- ´ based mmWave massive MIMO channel estimation problem. Our numerical results demonstrate the superiority of proposed method in terms of channel estimation accuracy and training overhead reduction.