GA-SMOTE-RF Enhanced Kalman Filter with Adaptive Noise Reduction.

Area of Science:

• Optical Engineering
• Signal Processing
• Machine Learning

Background:

• Atmospheric turbulence significantly degrades free-space laser communication (FSO) quality, limiting applications in military and rescue operations.
• Existing methods struggle to effectively classify interference and reduce noise in turbulent FSO channels.

Purpose of the Study:

• To develop an intelligent classification and adaptive noise reduction system for FSO communication.
• To significantly improve turbulence channel interference classification accuracy and overall communication quality.

Main Methods:

• Integration of genetic algorithms (GA), synthetic minority oversampling technique (SMOTE), random forest (RF), and Kalman filtering.
• Development of a hybrid approach for interference classification and adaptive noise reduction.

Main Results:

• Achieved 98.27% classification accuracy, 0.9732 F1-score, and 0.9653 MCC in simulations, outperforming SVM and KNN.
• Demonstrated low average RMSE (0.6983) and zero estimated delay in signal noise reduction.
• Validated practical applicability with real-world FSO data, achieving 0.308 RMSE and 0.75% Average Regret in Kalman filter parameter selection.

Conclusions:

• The proposed GA-SMOTE-RF system effectively mitigates atmospheric turbulence effects in FSO communication.
• The system offers excellent signal quality, efficient real-time processing, and practical applicability for real-world FSO systems.