Conventional empirical and deterministic models struggle in dense urban and mixed LOS/NLOS environments. Manual tuning is slow, and ray-based methods are computationally expensive. RFlex, our AI/ML based RF Planner, integrates Random Forest regression to learn real-world propagation behavior directly from data.
Random Forest is an ensemble of decision trees trained on different subsets of drive test and geospatial data. Each tree captures local propagation patterns, while the ensemble averages their outputs to deliver stable and high-accuracy predictions.
- Lower prediction error reduced MAE and RMSE compared to legacy models
- Robust urban performance even in complex NLOS scenarios
- Fast inference suitable for large-scale network simulations
- Data driven calibration using local drive test measurements
- Seamless integration with ray-based and empirical planning workflows
