Sum-Rate Maximization for FRIS-Enabled NOMA Communications

The fluid antenna system (FAS) concept introduces shape- and position-reconfigurability, unlocking significant performance gains for future wireless networks. In this letter, we extend it to fluid reconfigurable intelligent surfaces (FRIS), where each reflecting element can reposition itself on demand within a prescribed subregion. We investigate a multi-user FRIS-assisted downlink non-orthogonal multiple access (NOMA) system and formulate a network-level sum-rate maximization problem that jointly involves user clustering, FRIS element selection, phase-shift design, and NOMA power allocation under per-user quality-of-service (QoS) constraints. The resulting mixed-integer non-linear program is computationally challenging due to its hybrid binary-continuous decision space and the spatial correlation across the FRIS surface. To address this, we adopt a two-stage design: (i) distance-based interleaved clustering that partitions theUusers intoCNOMA clusters scheduled in orthogonal time slots, and (ii) per-cluster FRIS-NOMA optimization. For each cluster, we first apply a cross-entropy (CE) scheme to select a high-quality FRIS element-activation mask using a lightweight multi-user cascaded channel-gain metric, and then optimize the continuous phase shifts and NOMA power coefficients for the selected mask via a grey wolf optimizer (GWO) with a penalty-augmented fitness function. Simulation results demonstrate that the proposed FRIS-NOMA framework yields substantial sum-rate gains over conventional RIS-aided NOMA/OMA benchmarks and operates close to an ideal continuous-phase FRIS upper bound across diverse propagation conditions.