Path Loss Prediction Model of 5G Signal Based on Fusing Data and XGBoost-SHAP Method.

Area of Science:

• Telecommunications Engineering
• Wireless Communication Systems
• Signal Propagation Modeling

Background:

• Accurate path loss prediction is vital for optimizing wireless networks, especially 5G, which faces challenges from complex terrains and urban environments.
• Traditional models struggle with the nonlinearities and uncertainties of 5G signal propagation in areas with varied terrain and dense high-rise buildings.
• Existing methods often lack the precision needed for effective network planning and user experience enhancement in challenging environments.

Purpose of the Study:

• To develop innovative feature engineering and prediction models for accurate 5G signal path loss modeling in complex terrains.
• To overcome the limitations of traditional linear models in predicting 5G signal characteristics, particularly in mountainous regions.
• To enhance the accuracy and stability of 5G signal path loss predictions for improved wireless network planning.

Main Methods:

• Utilized smartphones as receivers to capture multimodal data, including 3D structures and obstructions, for 5G signal analysis in N1 and N78 bands.
• Implemented the XGBoost algorithm with Optuna for hyperparameter tuning to optimize prediction model performance.
• Employed SHAP values for interpreting the model's results and understanding the influence of environmental features on signal path loss.

Main Results:

• Achieved a breakthrough in 5G signal path loss prediction with a coefficient of determination (R²) of 0.76 and a Root Mean Square Error (RMSE) of 3.81 dBm.
• Demonstrated the effectiveness of the multimodal system and XGBoost model in handling complex environmental factors affecting signal propagation.
• Identified key environmental features impacting 5G signal path loss through SHAP value analysis.

Conclusions:

• The developed model significantly enhances the accuracy and stability of 5G signal path loss predictions in complex environments.
• This research provides a robust technical framework and theoretical foundation for optimizing wireless communication networks.
• The findings support improved planning and deployment strategies for future wireless communication systems in challenging terrains.