Integrating Over-the-Air Federated Learning and Non-Orthogonal Multiple Access: What Role Can RIS Play?

With the aim of integrating over-the-air federated learning (AirFL) and non-orthogonal multiple access (NOMA) into an on-demand universal framework, this paper proposes a novel reconfigurable intelligent surface (RIS)-aided hybrid network by leveraging RIS to flexibly adjust signal processing order heterogeneous data. The objective work is maximize achievable rate jointly optimizing transmit power, controlling receive scalar, designing phase shifts. Since concurrent transmissions all computation communication signals are aided discrete shifts at RIS, considered problem (P0) challenging mixed integer programming problem. To tackle intractable issue, we decompose original non-convex (P1) combinatorial (P2), which characterized continuous variables, respectively. For transceiver design (P1), power allocation subproblem first solved invoking difference-of-convex programming, then control addressed using successive convex approximation, where closed-form expressions simplified cases derived obtain deep insights. reflection relaxation-then-quantization method adopted find suboptimal solution for striking trade-off between complexity performance. Afterwards, alternating optimization algorithm developed solve non-linear iteratively. Finally, simulation results reveal that 1) proposed RIS-aided can support efficiently, 2) performance gains be improved properly selecting location 3) designed algorithms also applicable conventional networks with only AirFL or NOMA users.