AFDM-Aided Grouped RSMA for High-Mobility Downlink Transmissions

Abstract

Rate-splitting multiple access (RSMA) has emerged as a flexible downlink framework for managing multiuser interference through common and private message splitting. However, in high-mobility scenarios, Doppler-induced intercarrier interference degrades RSMA-based transmission and motivates integration with a transform-domain waveform. In this paper, we propose an affine frequency division multiplexing (AFDM)-aided grouped RSMA scheme with discrete affine Fourier transform (DAFT)-domain common-subspace design for high-mobility downlink transmissions. Unlike existing transform-domain RSMA designs, the proposed formulation treats common-stream decoding sets and group-common subspaces as DAFT-domain structural variables and explicitly accounts for the weakest-decoder common-rate bottleneck. Specifically, we first design a finite-candidate DAFT-domain structural selector to jointly determine user grouping and group-common subspaces according to DAFT-domain channel compatibility and common-rate performance constraints. Furthermore, a nested AFDM-structured uncertainty model is incorporated into the slow structural layer, ensuring that grouping, commonsubspace construction, and uncertainty-model selection are kept outside the per-slot continuous precoder optimization. With the selected structural variables fixed, the fast layer updates only the continuous reduced-space precoders through weighted minimum mean-square error (WMMSE)-based alternating optimization (AO). Simulation results show that the proposed scheme achieves higher continuous weighted sum-rate (WSR) and stronger Doppler stability than representative baselines, while incurring a reduced per-slot computational cost.