The Low-Complexity Design and Optimal Training Overhead for IRS-Assisted MISO Systems

A low-complexity channel estimation and passive beamforming design for intelligent reflecting surface (IRS)-assisted multiple-input single-output (MISO) systems is proposed. Specifically, we present the low-complexity framework for maximizing the achievable rate of IRS-assisted MISO systems with discrete phase shifters at each IRS element. In contrast to existing solutions, the training set of IRS reflection coefficient matrix is pre-designed and the effective superposition channel estimation and transmit beamforming design are then performed for each IRS reflection coefficient matrix in the training set. Following this, the IRS reflection optimization is simplified by selecting the one that maximizes the achievable rate from the pre-designed training set. Secondly, we analyze the theoretical performance of the proposed framework and provide the optimal training overhead for maximizing the effective achievable rate given the channel coherence time. Finally, numerical simulations evaluate the rate performance of the proposed design. In particular, simulation results demonstrate that the proposed framework is a competitive option in practical communication systems with channel estimation errors.