Spectral Efficiency Analysis and Optimization in Hybrid RIS Systems Under Nonlinear Power Amplifier Effects
摘要
Hybrid reconfigurable intelligent surfaces (RISs) enhance link budget by augmenting a large passive array with a small subset of actively amplified elements, but the power amplifiers (PAs) required for those active units introduce nonlinear distortion that can nullify the expected spectral-efficiency (SE) gain. This paper develops a Bussgang-based OFDM framework that captures the statistical impact of PA non-linearities and yields closed-form expressions for the received signal, the (per-subcarrier) SINR, and the resulting SE in both single- and multi-user downlinks. To mitigate distortion we cast the choice of active elements as a combinatorial minimization problem and study five variable complexity solvers: Greedy search, Random Search (RS), Genetic Algorithm (GA), Binary Particle Swarm Optimization (BPSO), and an Estimation-of-Distribution Algorithm (EDA). All population-based heuristics are tuned to a common evaluation budget, enabling a fair complexity–performance comparison. Numerical results show that GA, BPSO, EDA, and RS restore a monotonic SE increase with the number of active elements, approaching the ideal (linear) bound within