This repostiroy provides a reference implementation of Adaptive Optical Modulation enhanced with neural network prediction
The project demonstrates training and applying neural networks to dynamically predict and adjust optical modulation parameters, aiming to improve efficiency and accuracy over traditional methods.
All code here is original and provides a fully reproducible workflow for both the neural network training and the optical modulation algorithm.
This repository is intended for researchers, engineers, and enthusiasts interested in optical modulation, signal processing, and AI-enhanced control systems.
The method combines classical adaptive optical modulation with AI-based prediction to enable real-time optimization of optical signals.
This implementation is specifically oriented towards Free-Space Optics (FSO) applications, rather than fiber-optic systems, highlighting challenges such as atmospheric turbulence and line-of-sight variability.
While adaptive optical modulation has been studied previously, this repository provides a first open-source reference implementation integrating AI-based prediction.
All results, metrics, and examples are provided to facilitate understanding, reproduction, and extension of the method.
The concepts implemented in this repository were initially presented at The 2nd International Scientific and Technical Conference on "Infocommunication Systems and Artificial Intelligence Technologies" (December 4, 2024), held at the Azerbaijan Technical University. Although the work did not receive wide attention at the time, this repository represents a full, open implementation of those ideas.
The project is released under the AGPL-3.0 license, ensuring that any derivative work, service, or application built on this code must remain open-source and properly attribute the original source.
Contributions, experiments, or extensions of this method should maintain the open-source nature of the project.