Sparse channel estimation and passive beamforming with practical phase shift model for IRS-assisted OFDM systems

Abstract

Intelligent reflecting surfaces (IRS) enable control of wireless propagation environments and have emerged as a highly promising and cost-effective technology for elevating the performance of future wireless communication systems. Obtaining the passive beamforming gain requires accurate channel state information (CSI), despite the practical challenge posed by its passive reflecting elements, which lack transmitting or receiving capability. Prior research on IRS has mostly used an ideal phase shift model, which assumes that every element reflects the signal fully irrespective of its phase shift. However, achieving this in real-world situations is difficult. This paper proposes a practical phase-shift model that takes into account phase-dependent changes in amplitude within the subgrouping reflection coefficient. This paper presents a new formulation for sparse channel estimation and passive beamforming in IRS subgrouping-assisted orthogonal frequency division multiplexing (OFDM) systems, leveraging a practical phase-shift model. Channel sparsity in the time domain is exploited, taking into account the physical proximity of passive elements and utilizing common sparsity shared among different subgroups using the variational Bayesian inference (VBI) framework. Furthermore, we proposed a stochastic majorization-minimization (SMM) based approach to optimize the IRS phase shifts to maximize the achievable sum-rate for the system, accounting for the practical phase-shift model. The results of the simulation validate the efficiency of the proposed channel estimation and passive beamforming method, showing substantial performance enhancements compared to existing cutting-edge techniques.