Optimized RSMA and NOMA Transmission in MIMO Downlinks With Quantized Channel Direction Information

Channel state information is essential for optimizing transmit precoding in multiple-input multiple-output (MIMO) downlinks. This work assumes that only quantized channel direction information (CDI), representing the normalized channel vector, is available at the base station. With a finite number of CDI bits, we approximate the achievable rates for users in rate-splitting multiple access (RSMA) and non-orthogonal multiple access (NOMA) systems. Using these approximations, we propose a convex alternating optimization framework that incorporates geometric programming to achieve max-min rate fairness. The numerical results show significant rate gains from jointly optimizing transmit power, CDI rate, and common rate fractions in RSMA, as well as transmit power, CDI rate, and user pairing in NOMA. Our findings indicate that RSMA outperforms other schemes in high-SNR and low-to-moderate CDI rate regimes, while NOMA is most effective when pairing users with high channel correlation and large channel power differences.