Geometric Manifold Statistics of Turbulence-Impacted Beam Propagation and Compensation in Optical Communication

Abstract: The present study extends the analysis of turbulence-affected beam statistics through a manifold-based statistical framework that unifies probabilistic modeling with geometric interpretation. The spatial intensity distributions, distorted by dynamic turbulence, are represented using Gaussian Mixture Models (GMMs), whose probability landscapes are refined via Kernel Density Estimation (KDE) applied to pixel-level intensity data across temporal frames. The temporal evolution of turbulence is quantified by monitoring the variation of the unnormalized volumetric integrals under the GMM surfaces, providing a continuous measure of power redistribution within the optical field. Experimental investigations were carried out under four propagation conditions: turbulence-free reference, turbulence only, turbulence with a single PMMA compensator, and turbulence with dual PMMA compensators. To assess statistical dissimilarity, Affine-Invariant Riemannian Metric (AIRM) distances were computed between the covariance representations of successive frames, capturing the geometric evolution of beam topology in the space of Symmetric Positive Definite (SPD) matrices. Complementary topological distance analyses, performed with respect to the initial frame of each set, revealed distinct signatures of turbulence-induced aberrations and their progressive mitigation through dielectric compensation. The observed reduction in inter-frame distance metrics confirms the ability of coupled dipole interactions within PMMA slabs to partially restore statistical coherence in turbulence-distorted optical beams.