Fairness Driven Joint Phase and PAC Optimization for NOMA Transmission With D/ND-IRS

This paper addresses the problem of optimizing the rate performance in a downlink non-orthogonal multiple access (NOMA) system with an intelligent reflecting surface (IRS) capable of operating in diagonal (D) or non-diagonal (ND) configuration. The objective is to maximize the minimum signal-to-interference-plus-noise ratio (SINR), which is equivalent to maximizing the data rates for all users based on fairness. To support multiple users, a two-user pairing scheme is considered. Next, we propose a sub-optimal optimization (SO) approach to jointly optimize the phase shifts and power allocation coefficients. Further, we compare the data rate performance of different IRS configurations and user pairing schemes under various transmit signal-to-noise ratio (SNR) conditions. We propose a novel hybrid IRS configuration that selects diagonal or non-diagonal mode dynamically. Additionally, we compare our results with the optimal greedy search method. Our results demonstrate significant improvements in data rates using the hybrid IRS configuration with the SO approach (SO-hybrid) compared to traditional orthogonal multiple access (OMA) schemes for higher IRS elements ($M>64$). For instance, with two users, at an SNR of 20 dB and $M=225$, the SO-hybrid method achieves data rates up to 6.3 bps/Hz, compared to 5.3 bps/Hz and 5.4 bps/Hz for OMA with diagonal and non-diagonal IRS configurations, respectively. At 10 dB SNR, SO-hybrid reaches 4.3 bps/Hz, outperforming OMA-ND at 3.7 bps/Hz. Simulations further reveal the robustness of the hybrid NOMA scheme under imperfect successive interference cancellation and channel state information, proving its potential for enhancing wireless communication systems.