CPD-Based Channel Estimation for IRS-Assisted MIMO Communicaton

Abstract

Intelligent reflecting surface (IRS) have emerged as an innovative technology and are integrated into existing wireless communication systems for enhancing coverage and capacity. The acquisition of accurate channel state information is essential, but challenging due to the high dimensionality. Leveraging the derived tensor structure and uniqueness conditions for third-order CPD with a known factor matrix, a novel algebraic algorithm is proposed to solve the problem of channel estimation for IRS-assisted MIMO communication system. Moreover, a version of ALS that imposes ℓ2 regularisation on alternate optimization is proposed to obtain performance gains. Our proposed algebraic algorithm can be effective with low complexity at limited training overhead and large-scale IRS, which performs well in terms of estimation accuracy. Its output is valuable as an initialisation for iterative algorithms, especially in some demanding system settings. Numerical results demonstrated the effectiveness and usefulness of our proposed algorithms.