Performance Analysis of Multiple Reconfigurable Intelligent Surface-Assisted NOMA Networks

Abstract

Reconfigurable intelligent surfaces (RISs) can effectively enhance wireless coverage, yet a single RIS may be insufficient in blocked multi-user scenarios. This paper investigates a distributed multi-RIS-assisted downlink NOMA network with two representative transmission schemes. In comprehensive RIS-assisted NOMA (CRN), multiple RISs jointly assist transmission. In selective RIS-assisted NOMA (SRN), only the most suitable RIS is activated. Unlike idealized multi-RIS models, the proposed framework incorporates transmitter/receiver hardware impairments, RIS phase errors modeled by the von Mises distribution, imperfect channel state information (ipCSI), and imperfect successive interference cancellation (ipSIC). Under independent but not identically distributed Nakagami- $m$ cascaded channels, the end-to-end gains of CRN and SRN are characterized by moment-based Gamma approximations, from which tractable closed-form approximations for outage probability (OP) and ergodic capacity (EC) are derived. High-SNR asymptotic expressions and diversity-order results are also obtained, showing that imperfect CSI/SIC leads to OP floors, whereas practical impairments impose finite capacity ceilings at high SNR. An overhead-aware net energy-efficiency model is further developed by accounting for channel estimation, feedback/configuration, inter-RIS coordination, and RIS-selection overheads. Numerical results verify the analysis and reveal clear design tradeoffs: CRN achieves lower OP and higher EC, whereas SRN can be more energy efficient in low-rate or complexity-constrained regimes.