STAR-RIS Integrated Nonorthogonal Multiple Access and Over-the-Air Federated Learning: Framework, Analysis, and Optimization

This paper integrates non-orthogonal multiple access (NOMA) and over-the-air federated learning (AirFL) into a unified framework using one simultaneous transmitting reflecting reconfigurable intelligent surface (STAR-RIS). The STAR-RIS plays an important role in adjusting the decoding order of hybrid users for efficient interference mitigation omni-directional coverage extension. To capture impact non-ideal wireless channels on AirFL, closed-form expression optimality gap (a.k.a. convergence upper bound) between actual loss optimal is derived. analysis reveals that performance significantly affected by active passive beamforming schemes as well noise. Furthermore, when rate diminishes training proceeds, explicitly shown to converge with linear rate. accelerate while satisfying quality-of-service requirements, mixed-integer non-linear programming (MINLP) problem formulated jointly designing transmit power at configuration mode STAR-RIS. Next, trust region-based successive convex approximation method penalty-based semidefinite relaxation approach are proposed handle decoupled non-convex subproblems iteratively. An alternating optimization algorithm then developed find suboptimal solution original MINLP problem. Extensive simulation results show i) can efficiently support NOMA AirFL via concurrent uplink communications, ii) our algorithms achieve faster IID non-IID settings compared existing baselines, iii) both spectrum efficiency improved aid well-tuned