From Idealized Optical IRS Models to Realistic Lens-Based Architectures

We consider a lens-assisted intelligent reflecting surface (IRS) architecture for free-space optical (FSO) systems, where compact lens arrays focus incident light onto independently controlled IRS elements. Unlike existing models that assume a monolithic or coherently steered IRS, our structure introduces random angular variations across elements, leading to a fundamentally different channel behavior. We develop a novel statistical model to accurately capture this behavior using a single integral, despite the high dimensionality of the system. Analytical expressions for outage probability and BER are derived, incorporating turbulence, misalignment, and array geometry. Simulation results confirm the accuracy of the analysis and demonstrate the performance and robustness gains of the proposed architecture.