Fluid-Antenna-Aided SWIPT UAV–NOMA Relaying: Analysis and Deployment

Abstract

This paper develops an outage-analysis and deployment framework for a power-splitting simultaneous wireless information and power transfer (SWIPT)-enabled unmanned aerial vehicle (UAV) relay serving multiple fluidantenna users through half-duplex variable-gain amplify-and-forward nonorthogonal multiple access (NOMA) relaying. The instantaneous relay gain is derived from the harvested-energy constraint. Under the exact relay gain, each layer-decoding condition is transformed into a selected-gain-dependent threshold on the basestation–UAV channel, yielding an exact one-dimensional outage integral for an arbitrary distribution of the selected fluid-antenna gain. A relay-input signal-to-noise ratio (SNR) approximation then produces tractable correlated-port outage representations and an independent-port closed-form expression. Spatial dependence is modeled using a Gaussian copula calibrated to the power-gain correlation induced by a physical complexchannel model. The asymptotic analysis establishes diversity order one under a finite inverse-moment condition and separately characterizes the logarithmically modified single-port behavior. For deployment, a globally optimal bounded weighted-centroid benchmark is derived for a convex geometric surrogate, and a data-driven initializer validated using the exact outage objective is developed. The baseline derivations assume Rayleigh fading, linear energy harvesting, perfect channel state information, ideal successive interference cancellation, and a predetermined decoding order. Their robustness is numerically assessed under Rician fading, a sensitivityand-saturation nonlinear energy-harvesting model, aged port-selection channel state information, residual successive interference cancellation, and instantaneous order adaptation. The proposed placement initializer is also compared with random-forest and multilayer-perceptron alternatives.