Two-Timescale Beamforming Optimization for Intelligent Reflecting Surface Aided Multiuser Communication With QoS Constraints

Intelligent reflecting surface (IRS) is an emerging technology that able to reconfigure the wireless channel via tunable passive signal reflection and thereby enhance spectral energy efficiency of networks cost-effectively. In this paper, we study IRS-aided multiuser multiple-input single-output (MISO) system adopt two-timescale (TTS) transmission reduce processing complexity training overhead as compared existing schemes based on instantaneous state information (I-CSI), at same time, exploit diversity in scheduling. Specifically, long-term beamforming designed statistical CSI (S-CSI) all links, while short-term active cater I-CSI users' reconfigured channels with optimized IRS phase shifts. We aim minimize average transmit power access point (AP), subject individual quality service (QoS) constraints. The formulated stochastic optimization problem non-convex difficult solve since design variables are complicatedly coupled QoS To tackle problem, propose efficient algorithm, called primal-dual decomposition TTS joint (PDD-TJAPB), where original decomposed into a family problems, deep unfolding technique employed extract gradient from problems construct convex surrogate for problem. proposed algorithm proved converge stationary solution almost surely. Simulation results presented which demonstrate advantages effectiveness benchmark schemes.