Physics-informed adaptive transmission for coherent free-space optical communications: multi-dimensional amplitude-phase statistics.

Coherent free-space optical (FSO) communications suffer from coupled amplitude fading and phase distortion. Conventional adaptive modulation and coding (AMC) relying solely on signal-to-noise ratio (SNR) feedback cannot capture phase dynamics, leading to error floors under strong turbulence. We propose a physics-informed AMC framework that extracts scintillation index σI2 and phase variance σϕ2 from pilots. Combined with SNR, these form a multi-dimensional channel state vector for phase-aware adaptation without data-driven complexity. Modulation selection enforces turbulence-adaptive SNR and phase-variance thresholds, preventing high-order formats under severe phase distortion. Split-step Fourier simulations demonstrate error floor suppression, SNR gains of 0.6-2.0 dB at BER 10-3, 32-59% higher effective spectral efficiency, 80-88% of BICM capacity, and a decision processing latency as low as <22 µs, readily enabling real-time FPGA implementation.