Scalable IRS Panel Selection for UAV-Assisted Uplink Networks

Abstract

This paper investigates a UAV-assisted multi-panel intelligent reflecting surface (IRS) architecture to enhance link reliability and spectral efficiency in next-generation wireless networks. Unlike conventional IRS deployments with fixed geometries, UAV mobility introduces rapidly varying propagation conditions that challenge classical diversity and combining schemes. To address this, a Maximal Ratio Combining with Generalized Selection (MRC-GS) framework is proposed, in which the UAV optimizes its aerial position while IRS panels are selectively activated based on their contribution to the received signal. A joint optimization problem is formulated to design the IRS phase shifts, UAV position, and receiver combining weights. Hardware impairments are explicitly incorporated in the receiver combining design via impairments-aware noise covariance. The IRS phase optimization adopts a tractable effective-channel formulation and establishes the equivalence between maximizing the effective channel gain and signal-to-impairments-plus-noise ratio (SINR) under the adopted hardware impairment model. By activating only a dominant subset of IRS panels, the proposed approach improves energy efficiency and reduces computational overhead compared to full-activation schemes. Moreover, a low-complexity closed-form phase update is also provided for large-scale deployments. Simulation results demonstrate that the proposed MRC-GS-enabled UAV-IRS system reduces bit error rate (BER) and enhances spectral efficiency compared to benchmark single-IRS and full-combining schemes.