A Stacking-Based Method for Path Loss Prediction in Evaporation Ducts

Evaporation duct is a common atmospheric phenomenon in marine environments, which significantly affects signal transmission. Accurately predicting path loss (PL) in evaporation ducts is crucial for the design of maritime communication systems. This paper presents an ensemble method based on stacked generalization (stacking) to predict PL in evaporation ducts. Specifically, the stacking model employs a deep neural network (DNN) as the base learner in the first layer and a random forest as the meta learner in the second layer, using meteorological and antenna parameters as inputs to predict PL values. Compared to the state-of-the-art DNN model, the mean absolute error and mean square error values of the stacking model decrease by 41.57% and 20.81%, respectively. Meanwhile, the proportion of stacking model predictions with relative errors below 1% reaches 92.02%. Simulation results demonstrate that the stacking-based ensemble method outperforms single models and shows good potential for future investigations into evaporation ducts.