Tuning the parameters of a free-space optical channel using machine learning

Area of Science:

• Optical Communications
• Artificial Intelligence
• Machine Learning

Background:

• Free-space optical (FSO) technology offers high-speed data transmission but requires accurate performance prediction.
• Multiparametric simulations are crucial for understanding FSO system behavior under various conditions.

Purpose of the Study:

• To apply artificial intelligence (AI) and machine learning (ML) for simulating and predicting data transmission performance in FSO systems.
• To evaluate the effectiveness of different AI regression models in estimating the maximum quality factor (MaxQFactor) of FSO channels.

Main Methods:

• Utilized Optisystem software for multiparametric numerical simulations of FSO data transmission.
• Trained various AI models including Decision Tree Regression (DTR), Random Forest Regression (RFR), Gradient Boosting Regressor (GBR), Histogram Gradient Boosting Regressor (HGBR), and AdaBoost + Decision Tree Regression (ADDTR).
• Evaluated model performance using the coefficient of determination (R²), considering parameters like distance, attenuation, amplifier gains, beam divergence, and receiver diameter.

Main Results:

• DTR and RFR models achieved excellent prediction accuracy (R² > 95.00%) for the first simulation set (distance, attenuation, amplifier gains).
• DTR and RFR models also demonstrated excellent results (R² > 94.00%) in the second simulation set, incorporating beam divergence and receiver diameter.
• Graphical comparisons confirmed the high effectiveness of the AI methodology in predicting FSO channel output values.

Conclusions:

• AI and ML methodologies provide a highly effective and accurate approach for simulating and predicting FSO data transmission performance.
• The trained AI models, particularly DTR and RFR, can reliably estimate key performance metrics like MaxQFactor in FSO systems.
• This AI-driven simulation approach enhances the understanding and optimization of FSO communication systems.