Channel Estimation for IRS-Assisted Multi-User Millimeter Wave MIMO Systems

Abstract

Intelligent reflecting surface (IRS) is a promising technology for future wireless communication. To fully utilize the advantages of the IRS, the acquisition of accurate channel state information (CSI) is necessary. In this paper, we propose a manifold optimization-based channel estimation (MO-CE) algorithm for an IRS-assisted millimeter wave (mmWave) MIMO system with multiple users. In particular, by exploiting the sparsity of channel matrices and applying the Least Squares (LS) criterion, we formulate a multiple coupled complex fixed-rank (CFR) matrices optimization problem for channel estimation in the multiple users scenario. The problem is then transformed to several subproblems to reduce problem size and decouple multiple matrix variables. Furthermore, the CFR Hermitian positive semidefinite (PSD) manifold helps to satisfy CFR constraint by using semidefinite lifting. A Riemannian conjugate gradient-based movement is performed separately on the manifold, and the convergence of the proposed algorithm is established over the manifold too. Simulation results are provided to show the superiority of the proposed MO-CE algorithm in comparison with two benchmark schemes.