Performance of SWIPT for IRS-Aided Massive MIMO With Statistical CSI-Based Phase-Shift and Transmit Power Optimization

The performance of multi-user massive multiple-input multiple-output (MIMO) intelligent reflective surfaces (IRSs)-assisted simultaneous wireless information and power transfer (SWIPT) systems is investigated. By considering a hybrid time-switching/power-splitting protocol with a non-linear energy harvesting model, the average harvested energy, achievable rates, and energy-rate trade-offs are quantified. This analysis renders it useful for studying the impact of system imperfections and transmission impairments, including the erroneously estimated composite channels, spatial correlation, and statistical composite channel state information based user signal decoding, optimization of IRS phase-shifts, and transmit power control. Two strategies are adopted based on the max-min criterion and alternating procedures to optimize the corresponding energy-rate trade-offs while guaranteeing user-fairness. The asymptotically achievable harvested energy and user rates are derived in the high transmit power regime. A set of numerical results and Monte-Carlo simulations are provided to validate the convergence of optimization algorithms and theoretical performance characterizations. The performance gains and design insights specific to the proposed IRS-aided massive MIMO SWIPT system with statistical composite channel knowledge are also obtained through the numerical results.