Channel Estimation in IRS-assisted Multi-Cast Multi-group Communication Systems

The imperfection of channel state information (CSI) estimation in intelligent reflecting surface (IRS)-assisted multi-user systems may heavily reduce the network capacity. Therefore, in this paper, we first investigate the two-stage learning channel estimation (2S-CE) framework to enhance the accuracy of the three-phase channel estimation (3P-CE) algorithm in the literature. Then, we manage the actual transmitted data instead of sending all source data in IRSs-assisted multi-cast multi-group (IRS-MC-MG) systems to achieve high quality of service (QoS) and user satisfaction. Well-known zero-forcing (ZF) and block diagonalization (BD) techniques are adapted to achieve high-efficient solutions in IRS-MC-MG systems. Finally, we investigate adjustment algorithms to adapt to the environmental change and the channel estimation (CE) imperfection, thus can increase the received QoS and user satisfaction. Numerical results show our proposed framework decreases more than 38 times of error compared with the literature method, i.e., the 3P-CE algorithm.