• IRS-assisted RSMA systems demonstrate superior robustness over IRS-assisted SDMA under practical imperfections including hardware impairments, imperfect CSI, and discrete phase shifts.
• A sample average approximation (SAA)-based robust algorithm is proposed to jointly optimize IRS phase shifts and the beamforming matrix at the base station, effectively maximizing the weighted sum rate.
• The proposed SAA-based algorithm outperforms existing benchmark algorithms, confirming its effectiveness and robustness in realistic imperfect conditions.
• The system model explicitly captures the impacts of non-ideal hardware, imperfect CSI, and limited-resolution phase shifts, providing a comprehensive framework for robust design.