Deep Learning-based Joint Transmit and Reflective Beamforming Design for IRS-Aided MISO Multiuser Systems Under Statistical CSI

In this paper, we study the joint design of the transmit beamforming and reflective beamforming for an intelligent reflective surface (IRS)-aided multiple-input single-output (MISO) multiuser communication system. Particularly, we maximize the minimum achievable rate per user by jointly designing the phase shifts of IRS and active beamforming at the base station on the basis of the statistical channel state information (CSI). More important, we use the twin delayed deep deterministic policy gradient (TD3) algorithm with either traditional experience replay or priority experience replay (PER) to solve the optimization problem. Simulation results reveal that the TD3 achieves higher minimum average user data rate than the deep deterministic policy gradient algorithm. Additionally, the PER-TD3 algorithm based on statistical CSI has much lower computational complexity compared to the instantaneous one.