Optical wireless communication (OWC) is presented as a high-bandwidth, spatially secure, and RF-interference-free alternative for digital transmission. The technology's main weakness, however, is that free-space optical links depend heavily on an unobstructed line of sight.
The arXiv paper proposes using edge-AI to address this weakness. It describes a method for fault detection and bit-error rate diagnostics in optical wireless links, applying machine learning at the edge rather than in a central system.
The abstract does not include experimental results or implementation details. It does make clear that the motivation is the extreme sensitivity of direct free-space links to line-of-sight issues, and that edge-AI is being explored as a diagnostic tool for such links.