HHO and XGBoost-Based Intelligent Hybrid Data- and Model-Driven Maritime Path Loss Model

It is significant for next-generation maritime communication systems to predict maritime path loss accurately. To advance maritime wireless transmission services, this letter presents an intelligent-aided hybrid data- and model-driven maritime path loss model. The proposed framework combines Harris Hawks Optimization (HHO) with Extreme Gradient Boosting (XGBoost) to adaptively model path loss under the constraints of physical propagation theory across multiple frequencies. By leveraging intelligent optimization to guide the adjustment of physical model parameters, the approach significantly enhances modeling accuracy. Numerical results demonstrate that the extended model achieves an average Root Mean-Square Error (RMSE) of 1.76dB, corresponding to an 83.22% improvement over the ITU-R P.2001 model.